PACKET MULTI-CAST FOR MEMORY POOL REPLICATION
Examples described herein relate to a network interface. In some examples, the network interface is to access data designated for transmission in at least one packet to multiple memory nodes by inclusion of an multicast identifier of a memory node group and transmit the at least one packet to a destination network device, wherein the multicast identifier of the memory node group in the at least one packet is to cause an intermediate network device to multicast the packet to multiple memory nodes. In some examples, a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory. In some examples, the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
The present application claims the benefit of priority date of U.S. provisional patent application Ser. No. 63/115,511, filed Nov. 18, 2020, the entire disclosure of which is incorporated herein by reference.
Scale-out and distributed architectures increase computing resources or available memory or storage by adding processors, memory, and storage for access using a fabric or network. Disaggregated memory architectures can use pools of memory, located remote from the compute nodes in the system. A memory pool can be shared across a rack or set of racks in a data center. As the memory pools are remote from a compute node, there are additional latencies inherent in their accesses due to latency of packet formation and processing and media traversal across one or more network elements.
A data center comprising disaggregated resources, such as data center 100, can be used in a wide variety of contexts, such as enterprise, government, cloud service provider, and communications service provider (e.g., Telco's), as well in a wide variety of sizes, from cloud service provider mega-data centers that consume over 100,000 sq. ft. to single- or multi-rack installations for use in base stations.
The disaggregation of resources to sleds comprised predominantly of a single type of resource (e.g., compute sleds comprising primarily compute resources, memory sleds containing primarily memory resources), and the selective allocation and deallocation of the disaggregated resources to form a managed node assigned to execute a workload improves the operation and resource usage of the data center 100 relative to typical data centers comprised of hyperconverged servers containing compute, memory, storage and perhaps additional resources in a single chassis. For example, because sleds predominantly contain resources of a particular type, resources of a given type can be upgraded independently of other resources. Additionally, because different resources types (processors, storage, accelerators, etc.) typically have different refresh rates, greater resource utilization and reduced total cost of ownership may be achieved. For example, a data center operator can upgrade the processors throughout their facility by only swapping out the compute sleds. In such a case, accelerator and storage resources may not be contemporaneously upgraded and, rather, may be allowed to continue operating until those resources are scheduled for their own refresh. Resource utilization may also increase. For example, if managed nodes are composed based on requirements of the workloads that will be running on them, resources within a node are more likely to be fully utilized. Such utilization may allow for more managed nodes to run in a data center with a given set of resources, or for a data center expected to run a given set of workloads, to be built using fewer resources.
It should be appreciated that each of the other pods 120, 130, 140 (as well as any additional pods of the data center 100) may be similarly structured as, and have components similar to, the pod 110 shown in and described in regard to
Referring now to
In the illustrative embodiments, each sled of the data center 100 is embodied as a chassis-less sled. That is, each sled has a chassis-less circuit board substrate on which physical resources (e.g., processors, memory, accelerators, storage, etc.) are mounted as discussed in more detail below. As such, the rack 240 is configured to receive the chassis-less sleds. For example, each pair 310 of elongated support arms 312 defines a sled slot 320 of the rack 240, which is configured to receive a corresponding chassis-less sled. To do so, each illustrative elongated support arm 312 includes a circuit board guide 330 configured to receive the chassis-less circuit board substrate of the sled. Each circuit board guide 330 is secured to, or otherwise mounted to, a top side 332 of the corresponding elongated support arm 312. For example, in the illustrative embodiment, each circuit board guide 330 is mounted at a distal end of the corresponding elongated support arm 312 relative to the corresponding elongated support post 302, 304. For clarity of the Figures, not every circuit board guide 330 may be referenced in each Figure.
Each circuit board guide 330 includes an inner wall that defines a circuit board slot 380 configured to receive the chassis-less circuit board substrate of a sled 400 when the sled 400 is received in the corresponding sled slot 320 of the rack 240. To do so, as shown in
It should be appreciated that each circuit board guide 330 is dual sided. That is, each circuit board guide 330 includes an inner wall that defines a circuit board slot 380 on each side of the circuit board guide 330. In this way, each circuit board guide 330 can support a chassis-less circuit board substrate on either side. As such, a single additional elongated support post may be added to the rack 240 to turn the rack 240 into a two-rack solution that can hold twice as many sled slots 320 as shown in
In some embodiments, various interconnects may be routed upwardly or downwardly through the elongated support posts 302, 304. To facilitate such routing, each elongated support post 302, 304 includes an inner wall that defines an inner chamber in which interconnects may be located. The interconnects routed through the elongated support posts 302, 304 may be embodied as any type of interconnects including, but not limited to, data or communication interconnects to provide communication connections to each sled slot 320, power interconnects to provide power to each sled slot 320, and/or other types of interconnects.
The rack 240, in the illustrative embodiment, includes a support platform on which a corresponding optical data connector (not shown) is mounted. Each optical data connector is associated with a corresponding sled slot 320 and is configured to mate with an optical data connector of a corresponding sled 400 when the sled 400 is received in the corresponding sled slot 320. In some embodiments, optical connections between components (e.g., sleds, racks, and switches) in the data center 100 are made with a blind mate optical connection. For example, a door on each cable may prevent dust from contaminating the fiber inside the cable. In the process of connecting to a blind mate optical connector mechanism, the door is pushed open when the end of the cable approaches or enters the connector mechanism. Subsequently, the optical fiber inside the cable may enter a gel within the connector mechanism and the optical fiber of one cable comes into contact with the optical fiber of another cable within the gel inside the connector mechanism.
The illustrative rack 240 also includes a fan array 370 coupled to the cross-support arms of the rack 240. The fan array 370 includes one or more rows of cooling fans 372, which are aligned in a horizontal line between the elongated support posts 302, 304. In the illustrative embodiment, the fan array 370 includes a row of cooling fans 372 for each sled slot 320 of the rack 240. As discussed above, each sled 400 does not include any on-board cooling system in the illustrative embodiment and, as such, the fan array 370 provides cooling for each sled 400 received in the rack 240. Each rack 240, in the illustrative embodiment, also includes a power supply associated with each sled slot 320. Each power supply is secured to one of the elongated support arms 312 of the pair 310 of elongated support arms 312 that define the corresponding sled slot 320. For example, the rack 240 may include a power supply coupled or secured to each elongated support arm 312 extending from the elongated support post 302. Each power supply includes a power connector configured to mate with a power connector of the sled 400 when the sled 400 is received in the corresponding sled slot 320. In the illustrative embodiment, the sled 400 does not include any on-board power supply and, as such, the power supplies provided in the rack 240 supply power to corresponding sleds 400 when mounted to the rack 240. Each power supply is configured to satisfy the power requirements for its associated sled, which can vary from sled to sled. Additionally, the power supplies provided in the rack 240 can operate independent of each other. That is, within a single rack, a first power supply providing power to a compute sled can provide power levels that are different than power levels supplied by a second power supply providing power to an accelerator sled. The power supplies may be controllable at the sled level or rack level, and may be controlled locally by components on the associated sled or remotely, such as by another sled or an orchestrator.
Referring now to
As discussed above, the illustrative sled 400 includes a chassis-less circuit board substrate 602, which supports various physical resources (e.g., electrical components) mounted thereon. It should be appreciated that the circuit board substrate 602 is “chassis-less” in that the sled 400 does not include a housing or enclosure. Rather, the chassis-less circuit board substrate 602 is open to the local environment. The chassis-less circuit board substrate 602 may be formed from any material capable of supporting the various electrical components mounted thereon. For example, in an illustrative embodiment, the chassis-less circuit board substrate 602 is formed from an FR-4 glass-reinforced epoxy laminate material. Of course, other materials may be used to form the chassis-less circuit board substrate 602 in other embodiments.
As discussed in more detail below, the chassis-less circuit board substrate 602 includes multiple features that improve the thermal cooling characteristics of the various electrical components mounted on the chassis-less circuit board substrate 602. As discussed, the chassis-less circuit board substrate 602 does not include a housing or enclosure, which may improve the airflow over the electrical components of the sled 400 by reducing those structures that may inhibit air flow. For example, because the chassis-less circuit board substrate 602 is not positioned in an individual housing or enclosure, there is no vertically-arranged backplane (e.g., a backplate of the chassis) attached to the chassis-less circuit board substrate 602, which could inhibit air flow across the electrical components. Additionally, the chassis-less circuit board substrate 602 has a geometric shape configured to reduce the length of the airflow path across the electrical components mounted to the chassis-less circuit board substrate 602. For example, the illustrative chassis-less circuit board substrate 602 has a width 604 that is greater than a depth 606 of the chassis-less circuit board substrate 602. In one particular embodiment, for example, the chassis-less circuit board substrate 602 has a width of about 21 inches and a depth of about 9 inches, compared to a typical server that has a width of about 17 inches and a depth of about 39 inches. As such, an airflow path 608 that extends from a front edge 610 of the chassis-less circuit board substrate 602 toward a rear edge 612 has a shorter distance relative to typical servers, which may improve the thermal cooling characteristics of the sled 400. Furthermore, although not illustrated in
As discussed above, the illustrative sled 400 includes one or more physical resources 620 mounted to a top side 650 of the chassis-less circuit board substrate 602. Although two physical resources 620 are shown in
The sled 400 also includes one or more additional physical resources 630 mounted to the top side 650 of the chassis-less circuit board substrate 602. In the illustrative embodiment, the additional physical resources include a network interface controller (NIC) as discussed in more detail below. Of course, depending on the type and functionality of the sled 400, the physical resources 630 may include additional or other electrical components, circuits, and/or devices in other embodiments.
The physical resources 620 are communicatively coupled to the physical resources 630 via an input/output (I/O) subsystem 622. The I/O subsystem 622 may be embodied as circuitry and/or components to facilitate input/output operations with the physical resources 620, the physical resources 630, and/or other components of the sled 400. For example, the I/O subsystem 622 may be embodied as, or otherwise include, memory controller hubs, input/output control hubs, integrated sensor hubs, firmware devices, communication links (e.g., point-to-point links, bus links, wires, cables, waveguides, light guides, printed circuit board traces, etc.), and/or other components and subsystems to facilitate the input/output operations. In the illustrative embodiment, the I/O subsystem 622 is embodied as, or otherwise includes, a double data rate 4 (DDR4) data bus or a DDR5 data bus.
In some embodiments, the sled 400 may also include a resource-to-resource interconnect 624. The resource-to-resource interconnect 624 may be embodied as any type of communication interconnect capable of facilitating resource-to-resource communications. In the illustrative embodiment, the resource-to-resource interconnect 624 is embodied as a high-speed point-to-point interconnect (e.g., faster than the I/O subsystem 622). For example, the resource-to-resource interconnect 624 may be embodied as a QuickPath Interconnect (QPI), an UltraPath Interconnect (UPI), PCI express (PCIe), or other high-speed point-to-point interconnect dedicated to resource-to-resource communications.
The sled 400 also includes a power connector 640 configured to mate with a corresponding power connector of the rack 240 when the sled 400 is mounted in the corresponding rack 240. The sled 400 receives power from a power supply of the rack 240 via the power connector 640 to supply power to the various electrical components of the sled 400. That is, the sled 400 does not include any local power supply (e.g., an on-board power supply) to provide power to the electrical components of the sled 400. The exclusion of a local or on-board power supply facilitates the reduction in the overall footprint of the chassis-less circuit board substrate 602, which may increase the thermal cooling characteristics of the various electrical components mounted on the chassis-less circuit board substrate 602 as discussed above. In some embodiments, voltage regulators are placed on a bottom side 750 (see
In some embodiments, the sled 400 may also include mounting features 642 configured to mate with a mounting arm, or other structure, of a robot to facilitate the placement of the sled 600 in a rack 240 by the robot. The mounting features 642 may be embodied as any type of physical structures that allow the robot to grasp the sled 400 without damaging the chassis-less circuit board substrate 602 or the electrical components mounted thereto. For example, in some embodiments, the mounting features 642 may be embodied as non-conductive pads attached to the chassis-less circuit board substrate 602. In other embodiments, the mounting features may be embodied as brackets, braces, or other similar structures attached to the chassis-less circuit board substrate 602. The particular number, shape, size, and/or make-up of the mounting feature 642 may depend on the design of the robot configured to manage the sled 400.
Referring now to
The memory devices 720 may be embodied as any type of memory device capable of storing data for the physical resources 620 during operation of the sled 400, such as any type of volatile (e.g., dynamic random access memory (DRAM), etc.) or non-volatile memory. Volatile memory may be a storage medium that requires power to maintain the state of data stored by the medium. Non-limiting examples of volatile memory may include various types of random access memory (RAM), such as dynamic random access memory (DRAM) or static random access memory (SRAM). One particular type of DRAM that may be used in a memory module is synchronous dynamic random access memory (SDRAM). In particular embodiments, DRAM of a memory component may comply with a standard promulgated by JEDEC, such as JESD79F for DDR SDRAM, JESD79-2F for DDR2 SDRAM, JESD79-3F for DDR3 SDRAM, JESD79-4A for DDR4 SDRAM, JESD209 for Low Power DDR (LPDDR), JESD209-2 for LPDDR2, JESD209-3 for LPDDR3, and JESD209-4 for LPDDR4. Such standards (and similar standards) may be referred to as DDR-based standards and communication interfaces of the storage devices that implement such standards may be referred to as DDR-based interfaces.
In one embodiment, the memory device is a block addressable memory device, such as those based on NAND or NOR technologies. A block can be any size such as but not limited to 2 KB, 4 KB, 8 KB, and so forth. A memory device may also include next-generation nonvolatile devices, such as Intel Optane® memory or other byte addressable write-in-place nonvolatile memory devices. In one embodiment, the memory device may be or may include memory devices that use chalcogenide glass, multi-threshold level NAND flash memory, NOR flash memory, single or multi-level Phase Change Memory (PCM), a resistive memory, nanowire memory, ferroelectric transistor random access memory (FeTRAM), anti-ferroelectric memory, magnetoresistive random access memory (MRAM) memory that incorporates memristor technology, resistive memory including the metal oxide base, the oxygen vacancy base and the conductive bridge Random Access Memory (CB-RAM), or spin transfer torque (STT)-MRAM, a spintronic magnetic junction memory based device, a magnetic tunneling junction (MTJ) based device, a DW (Domain Wall) and SOT (Spin Orbit Transfer) based device, a thyristor based memory device, or a combination of any of the above, or other memory. The memory device may refer to the die itself and/or to a packaged memory product. In some embodiments, the memory device may comprise a transistor-less stackable cross point architecture in which memory cells sit at the intersection of word lines and bit lines and are individually addressable and in which bit storage is based on a change in bulk resistance.
Referring now to
In the illustrative compute sled 800, the physical resources 620 are embodied as processors 820. Although only two processors 820 are shown in
In some embodiments, the compute sled 800 may also include a processor-to-processor interconnect 842. Similar to the resource-to-resource interconnect 624 of the sled 400 discussed above, the processor-to-processor interconnect 842 may be embodied as any type of communication interconnect capable of facilitating processor-to-processor interconnect 842 communications. In the illustrative embodiment, the processor-to-processor interconnect 842 is embodied as a high-speed point-to-point interconnect (e.g., faster than the I/O subsystem 622). For example, the processor-to-processor interconnect 842 may be embodied as a QuickPath Interconnect (QPI), an UltraPath Interconnect (UPI), or other high-speed point-to-point interconnect dedicated to processor-to-processor communications (e.g., PCIe).
The compute sled 800 also includes a communication circuit 830. The illustrative communication circuit 830 includes a network interface controller (NIC) 832, which may also be referred to as a host fabric interface (HFI). The NIC 832 may be embodied as, or otherwise include, any type of integrated circuit, discrete circuits, controller chips, chipsets, add-in-boards, daughtercards, network interface cards, or other devices that may be used by the compute sled 800 to connect with another compute device (e.g., with other sleds 400). In some embodiments, the NIC 832 may be embodied as part of a system-on-a-chip (SoC) that includes one or more processors, or included on a multichip package that also contains one or more processors. In some embodiments, the NIC 832 may include a local processor (not shown) and/or a local memory (not shown) that are both local to the NIC 832. In such embodiments, the local processor of the NIC 832 may be capable of performing one or more of the functions of the processors 820. Additionally or alternatively, in such embodiments, the local memory of the NIC 832 may be integrated into one or more components of the compute sled at the board level, socket level, chip level, and/or other levels. In some examples, a network interface includes a network interface controller or a network interface card. In some examples, a network interface can include one or more of a network interface controller (NIC) 832, a host fabric interface (HFI), a host bus adapter (HBA), network interface connected to a bus or connection (e.g., PCIe, CXL, DDR, and so forth). In some examples, a network interface can be part of a switch or a system-on-chip (SoC).
The communication circuit 830 is communicatively coupled to an optical data connector 834. The optical data connector 834 is configured to mate with a corresponding optical data connector of the rack 240 when the compute sled 800 is mounted in the rack 240. Illustratively, the optical data connector 834 includes a plurality of optical fibers which lead from a mating surface of the optical data connector 834 to an optical transceiver 836. The optical transceiver 836 is configured to convert incoming optical signals from the rack-side optical data connector to electrical signals and to convert electrical signals to outgoing optical signals to the rack-side optical data connector. Although shown as forming part of the optical data connector 834 in the illustrative embodiment, the optical transceiver 836 may form a portion of the communication circuit 830 in other embodiments.
In some embodiments, the compute sled 800 may also include an expansion connector 840. In such embodiments, the expansion connector 840 is configured to mate with a corresponding connector of an expansion chassis-less circuit board substrate to provide additional physical resources to the compute sled 800. The additional physical resources may be used, for example, by the processors 820 during operation of the compute sled 800. The expansion chassis-less circuit board substrate may be substantially similar to the chassis-less circuit board substrate 602 discussed above and may include various electrical components mounted thereto. The particular electrical components mounted to the expansion chassis-less circuit board substrate may depend on the intended functionality of the expansion chassis-less circuit board substrate. For example, the expansion chassis-less circuit board substrate may provide additional compute resources, memory resources, and/or storage resources. As such, the additional physical resources of the expansion chassis-less circuit board substrate may include, but is not limited to, processors, memory devices, storage devices, and/or accelerator circuits including, for example, field programmable gate arrays (FPGA), application-specific integrated circuits (ASICs), security co-processors, graphics processing units (GPUs), machine learning circuits, or other specialized processors, controllers, devices, and/or circuits.
Referring now to
As discussed above, the individual processors 820 and communication circuit 830 are mounted to the top side 650 of the chassis-less circuit board substrate 602 such that no two heat-producing, electrical components shadow each other. In the illustrative embodiment, the processors 820 and communication circuit 830 are mounted in corresponding locations on the top side 650 of the chassis-less circuit board substrate 602 such that no two of those physical resources are linearly in-line with others along the direction of the airflow path 608. It should be appreciated that, although the optical data connector 834 is in-line with the communication circuit 830, the optical data connector 834 produces no or nominal heat during operation.
The memory devices 720 of the compute sled 800 are mounted to the bottom side 750 of the of the chassis-less circuit board substrate 602 as discussed above in regard to the sled 400. Although mounted to the bottom side 750, the memory devices 720 are communicatively coupled to the processors 820 located on the top side 650 via the I/O subsystem 622. Because the chassis-less circuit board substrate 602 is embodied as a double-sided circuit board, the memory devices 720 and the processors 820 may be communicatively coupled by one or more vias, connectors, or other mechanisms extending through the chassis-less circuit board substrate 602. Of course, each processor 820 may be communicatively coupled to a different set of one or more memory devices 720 in some embodiments. Alternatively, in other embodiments, each processor 820 may be communicatively coupled to each memory device 720. In some embodiments, the memory devices 720 may be mounted to one or more memory mezzanines on the bottom side of the chassis-less circuit board substrate 602 and may interconnect with a corresponding processor 820 through a ball-grid array.
Each of the processors 820 includes a heatsink 850 secured thereto. Due to the mounting of the memory devices 720 to the bottom side 750 of the chassis-less circuit board substrate 602 (as well as the vertical spacing of the sleds 400 in the corresponding rack 240), the top side 650 of the chassis-less circuit board substrate 602 includes additional “free” area or space that facilitates the use of heatsinks 850 having a larger size relative to traditional heatsinks used in typical servers. Additionally, due to the improved thermal cooling characteristics of the chassis-less circuit board substrate 602, none of the processor heatsinks 850 include cooling fans attached thereto. That is, each of the heatsinks 850 is embodied as a fan-less heatsink. In some embodiments, the heat sinks 850 mounted atop the processors 820 may overlap with the heat sink attached to the communication circuit 830 in the direction of the airflow path 608 due to their increased size, as illustratively suggested by
Referring now to
In the illustrative accelerator sled 1000, the physical resources 620 are embodied as accelerator circuits 1020. Although only two accelerator circuits 1020 are shown in
In some embodiments, the accelerator sled 1000 may also include an accelerator-to-accelerator interconnect 1042. Similar to the resource-to-resource interconnect 624 of the sled 600 discussed above, the accelerator-to-accelerator interconnect 1042 may be embodied as any type of communication interconnect capable of facilitating accelerator-to-accelerator communications. In the illustrative embodiment, the accelerator-to-accelerator interconnect 1042 is embodied as a high-speed point-to-point interconnect (e.g., faster than the I/O subsystem 622). For example, the accelerator-to-accelerator interconnect 1042 may be embodied as a QuickPath Interconnect (QPI), an UltraPath Interconnect (UPI), or other high-speed point-to-point interconnect dedicated to processor-to-processor communications. In some embodiments, the accelerator circuits 1020 may be daisy-chained with a primary accelerator circuit 1020 connected to the NIC 832 and memory 720 through the I/O subsystem 622 and a secondary accelerator circuit 1020 connected to the NIC 832 and memory 720 through a primary accelerator circuit 1020.
Referring now to
Referring now to
In the illustrative storage sled 1200, the physical resources 620 are embodied as storage controllers 1220. Although only two storage controllers 1220 are shown in
In some embodiments, the storage sled 1200 may also include a controller-to-controller interconnect 1242. Similar to the resource-to-resource interconnect 624 of the sled 400 discussed above, the controller-to-controller interconnect 1242 may be embodied as any type of communication interconnect capable of facilitating controller-to-controller communications. In the illustrative embodiment, the controller-to-controller interconnect 1242 is embodied as a high-speed point-to-point interconnect (e.g., faster than the I/O subsystem 622). For example, the controller-to-controller interconnect 1242 may be embodied as a QuickPath Interconnect (QPI), an UltraPath Interconnect (UPI), or other high-speed point-to-point interconnect dedicated to processor-to-processor communications.
Referring now to
The storage cage 1252 illustratively includes sixteen mounting slots 1256 and is capable of mounting and storing sixteen solid state drives 1254. Of course, the storage cage 1252 may be configured to store additional or fewer solid state drives 1254 in other embodiments. Additionally, in the illustrative embodiment, the solid state drivers are mounted vertically in the storage cage 1252, but may be mounted in the storage cage 1252 in a different orientation in other embodiments. Each solid state drive 1254 may be embodied as any type of data storage device capable of storing long term data. To do so, the solid state drives 1254 may include volatile and non-volatile memory devices discussed above.
As shown in
As discussed above, the individual storage controllers 1220 and the communication circuit 830 are mounted to the top side 650 of the chassis-less circuit board substrate 602 such that no two heat-producing, electrical components shadow each other. For example, the storage controllers 1220 and the communication circuit 830 are mounted in corresponding locations on the top side 650 of the chassis-less circuit board substrate 602 such that no two of those electrical components are linearly in-line with each other along the direction of the airflow path 608.
The memory devices 720 of the storage sled 1200 are mounted to the bottom side 750 of the of the chassis-less circuit board substrate 602 as discussed above in regard to the sled 400. Although mounted to the bottom side 750, the memory devices 720 are communicatively coupled to the storage controllers 1220 located on the top side 650 via the I/O subsystem 622. Again, because the chassis-less circuit board substrate 602 is embodied as a double-sided circuit board, the memory devices 720 and the storage controllers 1220 may be communicatively coupled by one or more vias, connectors, or other mechanisms extending through the chassis-less circuit board substrate 602. Each of the storage controllers 1220 includes a heatsink 1270 secured thereto. As discussed above, due to the improved thermal cooling characteristics of the chassis-less circuit board substrate 602 of the storage sled 1200, none of the heatsinks 1270 include cooling fans attached thereto. That is, each of the heatsinks 1270 is embodied as a fan-less heatsink.
Referring now to
In the illustrative memory sled 1400, the physical resources 620 are embodied as memory controllers 1420. Although only two memory controllers 1420 are shown in
In some embodiments, the memory sled 1400 may also include a controller-to-controller interconnect 1442. Similar to the resource-to-resource interconnect 624 of the sled 400 discussed above, the controller-to-controller interconnect 1442 may be embodied as any type of communication interconnect capable of facilitating controller-to-controller communications. In the illustrative embodiment, the controller-to-controller interconnect 1442 is embodied as a high-speed point-to-point interconnect (e.g., faster than the I/O subsystem 622). For example, the controller-to-controller interconnect 1442 may be embodied as a QuickPath Interconnect (QPI), an UltraPath Interconnect (UPI), or other high-speed point-to-point interconnect dedicated to processor-to-processor communications. As such, in some embodiments, a memory controller 1420 may access, through the controller-to-controller interconnect 1442, memory that is within the memory set 1432 associated with another memory controller 1420. In some embodiments, a scalable memory controller is made of multiple smaller memory controllers, referred to herein as “chiplets”, on a memory sled (e.g., the memory sled 1400). The chiplets may be interconnected (e.g., using EMIB (Embedded Multi-Die Interconnect Bridge)). The combined chiplet memory controller may scale up to a relatively large number of memory controllers and I/O ports, (e.g., up to 16 memory channels). In some embodiments, the memory controllers 1420 may implement a memory interleave (e.g., one memory address is mapped to the memory set 1430, the next memory address is mapped to the memory set 1432, and the third address is mapped to the memory set 1430, etc.). The interleaving may be managed within the memory controllers 1420, or from CPU sockets (e.g., of the compute sled 800) across network links to the memory sets 1430, 1432, and may improve the latency associated with performing memory access operations as compared to accessing contiguous memory addresses from the same memory device.
Further, in some embodiments, the memory sled 1400 may be connected to one or more other sleds 400 (e.g., in the same rack 240 or an adjacent rack 240) through a waveguide, using the waveguide connector 1480. In the illustrative embodiment, the waveguides are 64 millimeter waveguides that provide 16 Rx (e.g., receive) lanes and 16 Tx (e.g., transmit) lanes. Each lane, in the illustrative embodiment, is either 16 GHz or 32 GHz. In other embodiments, the frequencies may be different. Using a waveguide may provide high throughput access to the memory pool (e.g., the memory sets 1430, 1432) to another sled (e.g., a sled 400 in the same rack 240 or an adjacent rack 240 as the memory sled 1400) without adding to the load on the optical data connector 834.
Referring now to
Additionally, in some embodiments, the orchestrator server 1520 may identify trends in the resource utilization of the workload (e.g., the application 1532), such as by identifying phases of execution (e.g., time periods in which different operations, each having different resource utilizations characteristics, are performed) of the workload (e.g., the application 1532) and pre-emptively identifying available resources in the data center 100 and allocating them to the managed node 1570 (e.g., within a predefined time period of the associated phase beginning). In some embodiments, the orchestrator server 1520 may model performance based on various latencies and a distribution scheme to place workloads among compute sleds and other resources (e.g., accelerator sleds, memory sleds, storage sleds) in the data center 100. For example, the orchestrator server 1520 may utilize a model that accounts for the performance of resources on the sleds 400 (e.g., FPGA performance, memory access latency, etc.) and the performance (e.g., congestion, latency, bandwidth) of the path through the network to the resource (e.g., FPGA). As such, the orchestrator server 1520 may determine which resource(s) should be used with which workloads based on the total latency associated with each potential resource available in the data center 100 (e.g., the latency associated with the performance of the resource itself in addition to the latency associated with the path through the network between the compute sled executing the workload and the sled 400 on which the resource is located).
In some embodiments, the orchestrator server 1520 may generate a map of heat generation in the data center 100 using telemetry data (e.g., temperatures, fan speeds, etc.) reported from the sleds 400 and allocate resources to managed nodes as a function of the map of heat generation and predicted heat generation associated with different workloads, to maintain a target temperature and heat distribution in the data center 100. Additionally or alternatively, in some embodiments, the orchestrator server 1520 may organize received telemetry data into a hierarchical model that is indicative of a relationship between the managed nodes (e.g., a spatial relationship such as the physical locations of the resources of the managed nodes within the data center 100 and/or a functional relationship, such as groupings of the managed nodes by the customers the managed nodes provide services for, the types of functions typically performed by the managed nodes, managed nodes that typically share or exchange workloads among each other, etc.). Based on differences in the physical locations and resources in the managed nodes, a given workload may exhibit different resource utilizations (e.g., cause a different internal temperature, use a different percentage of processor or memory capacity) across the resources of different managed nodes. The orchestrator server 1520 may determine the differences based on the telemetry data stored in the hierarchical model and factor the differences into a prediction of future resource utilization of a workload if the workload is reassigned from one managed node to another managed node, to accurately balance resource utilization in the data center 100. In some embodiments, the orchestrator server 1520 may identify patterns in resource utilization phases of the workloads and use the patterns to predict future resource utilization of the workloads.
To reduce the computational load on the orchestrator server 1520 and the data transfer load on the network, in some embodiments, the orchestrator server 1520 may send self-test information to the sleds 400 to enable each sled 400 to locally (e.g., on the sled 400) determine whether telemetry data generated by the sled 400 satisfies one or more conditions (e.g., an available capacity that satisfies a predefined threshold, a temperature that satisfies a predefined threshold, etc.). Each sled 400 may then report back a simplified result (e.g., yes or no) to the orchestrator server 1520, which the orchestrator server 1520 may utilize in determining the allocation of resources to managed nodes.
Packet Multi-CastWhen memory is disaggregated over a fabric or network and available in memory pools, a memory node can become a single point of failure for multiple servers that have pages of data residing on that memory node. If a memory node fails, all data on that node may be lost. To address this potential loss of data, data can be replicated or striped in one or more other nodes so that data can be recovered. A second memory node can be added, and the servers replicate the data to both memory nodes. However, for a single memory write, this can result in more fabric bandwidth consumed on the host link, because the host sends two copies of the same data to two different nodes.
Some solutions provide that a host issue multiple writes to different storage devices with different destination addresses by multicast. However, a write operation is not considered finished until all multicast operations are completed and the application issuing the write operation stalls, as dependent operations after that write operation cannot execute until the write operation completes. Some solutions provide that a memory node duplicate data into one or more other peer storage nodes instead of the host issuing multiple write operations. However, some solutions can introduce an extra network element or device traversal (e.g., hop) for the data and may increase latency of completion of a write operation because the data is written to one memory node and then replicated to another memory node.
Various embodiments use a variation of multicast where the memory nodes subscribe to a multi-cast address and the server sends a memory write to a network device (e.g., switch) with an multi-cast address (e.g., mcast) address. The host can send the data only once and the network device can replicate the data and transmit the data to multiple destination memory nodes. Various embodiments can reduce fabric bandwidth utilization by approximately 50% and allow approximately a two times improvement in memory write performance for disaggregated memory. By freeing up bandwidth usage on the links used by data replication operations, read performance can be improved and network or fabric congestion lowered. In addition, jitter (e.g., variation in packet latency from end-to-end in a network) may be reduced.
Multicast operations on a network fabric that uses Ethernet (e.g., IEEE 802.3-2018 and predecessors, revisions or variations thereof) can be unreliable. Various embodiments provide a manner of managing packet re-transmissions based on receipt of acknowledgements from all memory nodes in a group as described herein.
High speed interconnects can be used to couple a compute node 1600 and memory nodes 1650-0 to 1650-M-1 such as one or more of: Ethernet (IEEE 802.3), remote direct memory access (RDMA), InfiniBand, Internet Wide Area RDMA Protocol (iWARP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), quick UDP Internet Connections (QUIC), RDMA over Converged Ethernet (RoCE), Peripheral Component Interconnect express (PCIe), Intel QuickPath Interconnect (QPI), Intel Ultra Path Interconnect (UPI), Intel On-Chip System Fabric (IOSF), Omnipath, Compute Express Link (CXL), HyperTransport, high-speed fabric, NVLink, Advanced Microcontroller Bus Architecture (AMBA) interconnect, OpenCAPI, Gen-Z, Cache Coherent Interconnect for Accelerators (CCIX), Infinity Fabric (IF), 3GPP Long Term Evolution (LTE) (4G), 3GPP 5G, and variations thereof.
Compute node 1600 can include a network or fabric interface, compute resources (e.g., CPUs or accelerators), and memory resources (e.g., memory, storage, or cache). For example, compute node 1600 can execute workloads or applications. For example, compute node 1600 can be implemented as a server, rack of servers, computing platform, or others. In some examples, compute node 1600 can include one or more of: a core, graphics processing unit (GPU), field programmable gate array (FPGA), or application specific integrated circuit (ASIC). In some examples, a core can be sold or designed by Intel®, ARM®, AMD®, Qualcomm®, IBM®, Texas Instruments®, among others. Any processor can execute an operating system, driver, applications, and/or a virtualized execution environment (VEE) (e.g., virtual machine or container). In some examples, an operating system (OS) can be Linux®, Windows®, FreeBSD®, Android®, MacOS®, iOS®, or any other operating system. Compute node 1600 can be coupled to memory pool node 1650 using a network or fabric.
Any of memory pools 1650-0 to 1650-M-1 can include a network interface 1652, CPUs 1654, memory pool 1656 including any type of cache, volatile memory, non-volatile memory. Some embodiments of memory pools 1650-0 to 1650-M-1 can include accelerator devices, GPUs, or any component in any compute node 1600. Note that in some examples, any of memory pools 1650-0 to 1650-M-1 could also utilize a DMA engine and network interface so that data transfers from a memory pool to another memory pool, data transfers within or among one or more memory devices of a memory pool, or from a memory pool to a compute node could be initiated using a DMA transfer request. Any memory pool node 1650 can include a network or fabric interface, compute resources (e.g., CPUs or accelerators), and memory resources (e.g., memory, storage, or cache).
A memory hierarchy can be composed using any register, cache, local memory device or memory device in a memory pool node. Various embodiments can use at least two levels of memory hierarchy. Local memory 1608 can provide a first level of memory (alternatively referred to herein as “near memory”) including smaller faster memory made of, for example, DRAM or other volatile memory. Any memory device in any of memory pool nodes 1650-0 to 1650-M-1 can provide a second level or third level (or higher numbered level) which includes larger and slower (with respect to the near memory) volatile memory (e.g., DRAM) or nonvolatile memory storage (e.g., flash memory or byte addressable non-volatile memory (e.g., Intel® Optane® or Samsung Z-NAND)). The far memory can be presented as “main memory” to the host operating system (OS), while the near memory is a cache for the far memory that is transparent to the OS, thus rendering the embodiments described below to appear the same as prior art main memory solutions.
In some examples of a memory pool node, a CPU 1654 can be part of a SoC and network interface 1652 can access a memory device using an uncore without involving the SoC or CPU 1654. In some examples of a memory pool node, network interface 1652 can could interact with a memory controller to perform read or write memory devices in a memory pool 1656.
Various embodiments of DMA engine 1602 can be accessed by a requester using DMA semantics. In some examples, a DMA data transfer request can be specified using the following format:
Data_transfer(source address, destination address, length).
DMA engine 1602 could inspect address information (e.g., a physical or virtual address, address space identifier, etc.) in the DMA data transfer request and use that information to determine where the source or destination data is located (e.g., in local memory 1608 or in remote pool memory of some memory pool attached to the overall system via one or more switches or routers). When the address information indicates source or destination is remote, DMA engine 1602 can use network interface 1604 to carry messages to the remote memory pool to facilitate a read or write of pool memory. The messaging can support read or write semantics with appropriate reliability guarantees such as packet receipt Acknowledgements or negative-acknowledgement (NACK), reliable transport protocol (RTP), Cyclic Redundancy Check (CRC), checksums, re-transmit, packet sequence numbering, remote direct memory access (RDMA), UDP, and so forth. In some examples, DMA engine 1602 and/or network interface 1604 can be formed in a system on a chip (SOC) with one or more of CPUs 1606 and local memory 1608. In some examples, DMA engine 1602 and/or network interface 1604 can be coupled to one or more of CPUs 1606 and local memory 1608 using a device interface (e.g., DDR, CXL, PCIe). In some examples, network interface 1604 can directly access local memory 1608 without intervention by CPUs 1606 to access data.
DMA is a technology that allows an input/output (I/O) device to bypass a central processing unit (CPU) or core, and to send or receive data directly to or from a system memory. Because DMA allows the CPU or core to not manage a copy operation when sending or receiving data to or from the system memory, the CPU or core can be available to perform other operations. Without DMA, when the CPU or core is using programmed input/output, the CPU or core is typically occupied for the entire duration of a read or write operation and is unavailable to perform other work. With DMA, the CPU or core can, for example, initiate a data transfer, and then perform other operations while the data transfer is in progress. The CPU or core can receive an interrupt from a DMA controller when the data transfer is finished.
Some examples of DMA engine 1602 provide data mover and transformation operations. For example, DMA engine 1602 can validate CRC or checksum values in connection with storage and networking applications. For example, DMA engine 1602 can perform memory compare and delta generation or merge to support VM or container migration, VM or container check-pointing (e.g., to revert a VM or container to a previous state) and software managed memory deduplication usages.
Some examples of DMA engine 1602 are part of an Infrastructure Processing Unit (IPU). An IPU can include a SmartNIC with one or more programmable or fixed function processors to perform offload of operations that could have been performed by a CPU. The IPU can include one or more memory devices. In some examples, the IPU can perform virtual switch operations, manage storage transactions (e.g., compression, cryptography, virtualization), and manage operations performed on other IPUs, servers, or devices.
In some examples, a DMA command can zero-fills a region (e.g., write all zeros to a region of memory addresses) or a DMA command could cause compression of data in a region. For example, a DMA command could include fields such as [operation, address, parameter], where an operation could include zero fill or compression.
Local memory 1608 can include one or more of: one or more registers, one or more cache devices (e.g., level 1 cache (L1), level 2 cache (L2), level 3 cache (L3), lower level cache (LLC)), volatile memory device, non-volatile memory device, or persistent memory device. For example, memory 1608 can include static random access memory (SRAM) memory technology or memory technology consistent with high bandwidth memory (HBM), or double data rate (DDR), among others. Memory 1608 can be connected to network interface 1604 using a high speed interface (e.g., DDR, CXL (e.g., Compute Express Link Specification revision 2.0, version 0.9 (2020), as well as earlier versions, revisions or variations thereof), Peripheral Component Interconnect express (PCIe) (e.g., PCI Express Base Specification 1.0 (2002), as well as earlier versions, revisions or variations thereof).
A request can be issued by an application, VEE or other software executed by CPUs 1606 or other device such as an accelerator device. In some examples, in compute node 1600, DMA engine 1602 can determine whether the source address corresponds to an address in local memory 1608 or any of memory pools 1650-0 to 1650-M-1. To identify which addresses correspond to local memory 1608 or to devices in memory pool 1650-0 to 1650-M-1, DMA engine 1602 can be configured by one or more of: a management entity, orchestrator (e.g., Kubernetes, OpenStack, Slurm (High Performance Computing (HPC)), Open Source NFV Management and Orchestration (MANO) from European Telecommunications Standards Institute (ETSI)'s Open Source Mano (OSM) group), system software running on a local node (e.g., directed by a global orchestrator), pod manager, or traffic manager on a same or different host node, fabric manager (e.g., CXL fabric manager).
For example, a node 1 can build an overall memory layout across the system with nodes 1-4 whereby node 2 is allocated 3 GB from memory pool node 3, node 4 is allocated 2 GB from memory pool node 3. The orchestrator can coordinate with system software on each node to configure their state appropriately. For example, node 1 can inform node 2 that node 2 has 3 GB from node 3 which is at IP 10.10.10.1 and node 2 then configures its hardware state to reflect the configuration.
For example, DMA engine 1602 can utilize a conversion table to determine whether a source address corresponds to an address in local memory 1608 or any of memory pools 1650-0 to 1650-M-1. For example, the following can represent a conversion of an address to a memory device (e.g., local or remote).
There can be an arbitrary mapping of addresses to memory devices or IP addresses. A contiguous memory range mapping can be used. For example, a range from a set of smaller, contiguous ranges could be available for allocation with discontinuities. For example, memory pool can be associated with at addresses {0x1000-0x2000, 0x4000-0x5000, 0x7000-0x8000}.
In cases where the source address and destination address correspond to memory regions in local memory 1608, DMA engine 1602 can perform a copy of content in local memory 1608 starting at the source address and up to the specified length to a memory region in local memory 1608 starting at the destination address.
In cases where the source address corresponds to memory regions in local memory 1608 but the destination address corresponds to memory regions in any of memory pools 1650-0 to 1650-M-1, DMA engine 1602 can perform a copy of content in local memory 1608 starting at the source address and up to the specified length and utilize network interface 1604 to copy the content to a memory region corresponding to the destination address in one or more of memory pools 1650-0 to 1650-M-1. Some examples use RDMA or NVMe-oF protocols to copy content to remote memory.
In cases where the source address corresponds to a memory region in any of memory pools 1650-0 to 1650-M-1, but the destination address corresponds to a memory region in local memory 1608, DMA engine 1602 can utilize network interface 1604 to request a copy of content in a memory region corresponding to the destination address and length in one or more of memory pools 1650-0 to 1650-M-1. DMA engine 1602 can copy the received content into local memory 1608. Some examples use RDMA or NVMe-oF protocols to access content from remote memory.
In cases where the source and destination addresses correspond to memory regions in any of memory pools 1650-0 to 1650-M-1, DMA engine 1602 can utilize network interface 1604 to request a memory pool to perform a copy operation of content to another memory pool.
If a source address or destination address is in any of memory pools 1650-0 to 1650-M-1, DMA engine 1602 can generate one or more packets with a memory request. Where a source address corresponds to a memory device in any of memory pools 1650-0 to 1650-M-1, the packets can convey a memory request with a read request and the source address and length. Where a destination address corresponds to a memory device in any of memory pools 1650-0 to 1650-M-1, the packets can convey a memory request with a write request and the destination address and length. A memory pool can respond to a read request by providing the requested content, where the request is authorized. A memory pool can respond to a write request by providing an acknowledgement that the write request was completed, where the request is authorized.
If a source address or destination address is a virtual address, DMA engine 1602 can perform virtual address translation to generate a physical address corresponding to the virtual address. Some embodiments utilize virtual address translation in a DMA engine for remote data access without the requester managing network buffers. For example, a requester can access local buffers in local memory 1608 and not manage network connections, setup or shutdown. DMA engine 1602 can manage virtual addresses and translation
In some examples, to communicate with any of memory pools 1650-0 to 1650-M-1, network interface 1604 can utilize a remote direct memory access (RDMA) protocol with gets and puts or reads or writes. Examples of RDMA protocols include InfiniB and, Internet Wide Area RDMA Protocol (iWARP), quick UDP Internet Connections (QUIC), RDMA over Converged Ethernet (RoCE). In some examples, RDMA-enabled network interface is managed by requester software by management of its network construct, establishing a connection and managing network events. Some embodiments allow a requester to merely issue a DMA request and DMA engine 1602 and network interface 1604 handle setting up a connection, connection management, and RDMA setup with a target memory node. Accordingly, a requester may not need to initiate connection set up or perform a packet transmission but can merely issue a DMA request. The requester can specify source and destination address and length to DMA engine 1602 through a driver or library.
For example, for a put or get operation, network interface 1604 can be RDMA-enabled to setup remote RDMA operations including send queues and receive queues. RDMA can involve direct writes or reads to copy content of buffers across a connection without the operating system managing the copies. A network interface or other interface to a connection can implement a direct memory access engine and create a channel from its RDMA engine though a bus to application memory. A send queue and receive queue can be used to transfer work requests and are referred to as a Queue Pair (QP). A requester can place work request instructions on its work queues that tells the interface contents of what buffers to send to or receive content from. A work request can include an identifier (e.g., pointer or memory address of a buffer). For example, a work request placed on a send queue (SQ) can include an identifier of a message or content in a buffer (e.g., app buffer) to be sent. By contrast, an identifier in a work request in a Receive Queue (RQ) can include a pointer to a buffer (e.g., app buffer) where content of an incoming message can be stored. An RQ can be used to receive an RDMA-based command or RDMA-based response. A Completion Queue (CQ) can be used to notify when the instructions placed on the work queues have been completed.
For a get or read operation, network interface 1604 can send one or more packets with a get or read operation, with source address (e.g., translated to a physical address or not translated to a physical address) and length, to a target memory node. Network interface 1604 can copy the received data from a memory node to local memory 1608.
For example, a “pull” model can be used whereby a compute node can send “put from address x” to a memory pool and memory pool responds with “get from address x” to compute node and compute node responds with data to memory pool.
In some examples, system software can move or migrate data between a remote memory pool node and local memory to store more frequently accessed data in local memory 1608 and less frequently accessed data in a remote memory pool node. For data movement, various embodiments operate at a page granularity (e.g., application virtual memory managed by an operating system). A page in a virtual memory sense can refer to an underlying page size that the virtual memory supports. In some examples, for transfers of data from local memory 1710 to a memory pool node, or vice versa, system software (e.g., Operating system or hypervisors such as Linux, VMware ESX, Windows Hyper-V) could generate a DMA data transfer request to DMA engine 1602. Such operations could appear to DMA engine 1602 as requests for DMA transfers. However, either the source or destination address could be resident in a remote memory pool. In this approach, then, a unified API could be provided for moving data between local memory and remote memory (or among remote memory nodes).
System software running on compute node 1600 can use a page-based approach to manage an overall memory space split between local memory and pool memory in some set of memory pool nodes reachable from compute node 1600.
For redundant data copies across multiple memory nodes, various embodiments apply multicasting to a group of two or more memory nodes in a memory pool and enhance multicasting with reliability using a higher-level memory-over-fabric protocol described herein. Various embodiments use IP multicasting or Ethernet multicast (layer 2) or any other fabric specific multicast technology. In some examples, multicast over QUIC can be used consistent at least with Hypertext Transfer Protocol (HTTP) over multicast QUIC, draft-pardue-quic-http-mcast-07 (2020).
For example, any host node can issue a memory or storage access transaction to a pooled memory adapter (PMA) device 1662 using a device interface compatible for example with peripheral components interconnect express (e.g., PCI Express® Base Specification Revision 3.0 (2010) as well as earlier versions thereof, revisions and modifications thereof) or using a memory interface, or DMA command, as described herein. In some examples, a memory or storage access request can include: read type/write type, address (e.g., virtual address), read-modify-write, and data for a write request. PMA 1662 connected to a host 1660 can translate a host specific memory request to a fabric specific operation that enables it to be transported to a target memory node 1680. For example, based on an address provided with a memory or storage access request, a PMA 1662 can determine a destination IP address of a switch that is to manage memory access operations to the destination memory node. For example, PMA 1662 can be configured to associate a destination IP address of a particular switch with a memory address associated with a memory or storage access request. PMA 1662 can generate and transmit one or more packets to or with a multi-cast address and switch 1670 determines that the destination memory nodes associated with the multi-cast address.
If the data is to be replicated (e.g., as configured by a management entity) and transmitted to multiple memory nodes, the source PMA 1662 can transmit the write data multiple times, each time encapsulating the data with the appropriate memory node's network address. This can result in the data traversing the host to switch segment multiple times, and using valuable bandwidth on that link.
Source PMA 1662 can include a network interface card, network interface controller, network interface device or other communications device that can perform header and protocol processing of received packets. Source PMA 1662 can include any implementation of a network interface or DMA engine described herein. For communications over a media using Ethernet compatible packets, the host PMA can encapsulate the memory request with a network routable header such as a User Datagram Protocol (UDP)/internet protocol (IP)/Ethernet datagram that includes a destination network address of the memory node. For example, a memory store request can be translated to a memory write command, and along with the data, can be encapsulated with an Ethernet/IP header to be transmitted on a wired or wireless medium for a memory-over-fabric protocol. Memory access requests sent to switch 1670 can include a load or store (e.g., read or write) of an N-byte payload (where N is an integer) to a memory node as well as the accompanying data payload. In some examples, source PMA 1662 and switch 1670 are coupled using wired or wireless media with zero or more intervening network elements. In some examples, switch 1670 and receiver PMA 1682 can be coupled using wired or wireless media with zero or more intervening network elements.
A PMA 1662 can copy or store data to virtualized storage nodes using a protocol such as non-volatile memory express (NVMe) or NVMe over Fabrics (NVMe-oF). For example, NVMe-oF is described at least in NVM Express, Inc., “NVM Express Over Fabrics,” Revision 1.0, Jun. 5, 2016, and specifications referenced therein and variations and revisions thereof. In some examples, a remote memory node 1680 (or remote storage device) can communicate with a host node 1660 (e.g., host, server, or other remote memory pool or storage device) using at least one forwarding network device (e.g., switch) or a network cable. In some examples, a remote memory pool or remote storage device and an issuer of a memory access request are not located on a same system on chip (SoC).
Any host node 1660 can utilize one or multiple groups of memory nodes. In some examples, switch 1670 can route memory access requests from PMA 1662 to one or more memory nodes in a node group. For example, based on a memory address provided with a memory or storage access request, switch 1670 can determine a destination IP address of one or more memory nodes in a node group. For example, switch can be configured to associate destination IP addresses of memory nodes in a node group with a memory address associated with a memory or storage access request. A fabric switch (or a complex of switches) can connect host nodes with the memory nodes.
For a write operation from a host node 1660, a source PMA 1662 can decide which replica or node group to use for this write (e.g., as configured by a management entity) and identify a multicast group network address assigned to this group to encapsulate the data and transmit the data to switch 1670. In this example, a single switch 1670 is shown, but multiple switches can be used and one or more of the switches can be configured to perform multi-cast in accordance with embodiments described herein. According to some embodiments, a source PMA 1662 can include a multicast identifier (ID) with a packet to switch 1670 and switch 1670 can send copies of the packet to multiple memory nodes of a group associated with the multicast ID. In some examples, the multicast ID can be includes in a field in a UDP header, IP header, or other header. Switch 1670 can replicate a packet with a multicast ID and send the packet to memory nodes of a group. For example, switch 1670 can be configured with a look-up-table that associates multicast ID with destination IP addresses of memory nodes of a group. An example look-up-table is shown below.
In some examples, a host node 1660 can cause multicast of modified data (e.g., dirtied data) corresponding to a memory address associated with a device in a memory pool. For example, if data stored in a cache of host 1660 or source PMA 1662 is modified or updated, the data can be multicast to multiple memory nodes in accordance with embodiments described herein.
In some examples, in a similar manner as Redundant Array of Inexpensive Disks (RAID) (e.g., RAID 0 to RAID 6), data can be mirrored or striped across one or multiple memory nodes using multicast techniques described herein. Striping of data can include spreading data across multiple memory nodes. In some examples, RAID parity data calculated across data can be sent to one memory node or multicast to multiple memory nodes using multicast techniques described herein. In some examples, memory nodes can include hard disks or solid state drives.
A receiver PMA 1682 can include a network interface card, network interface controller, network interface device or other communications device that can perform header and protocol processing of received packets. A receiver PMA 1682 can include any implementation of a network interface or DMA engine described herein. A receiver PMA 1682 can form a memory access request from one or more received packets. A receiver PMA 1682 can be used by a memory node 1680 to translate fabric operations to a memory operation for the memory node. A memory node 1680 can include a combination of various volatile and non-volatile memory or storage technologies described herein as well as variations thereof. A memory node 1680 can include one or more tiers of memory. For example, multiple tiers of memory or storage can include one or more of: one or more registers, one or more cache devices (e.g., level 1 (L1), level 2 (L2), level 3 (L3), lower level cache (LLC)), memory (e.g., JEDEC DDR compatible memory or JEDEC HBM compatible memory), persistent memory (e.g., Intel® Optane® or Samsung Z-NAND), storage (e.g., NAND or 3D NAND), hard disk drives, or 2-level memory (2LM).
One or more host nodes can utilize a unique multicast group. However, a memory node can be part of any or all multicast groups, allowing each host to send memory access requests to all the memory nodes. Security policies can constrain this general solution if needed to limit access to specific memory nodes. Further, assuming n-way replication, each n-tuple memory nodes along with a host node can be assigned their own multicast group. This may allow a host per memory operation level granularity to replicate data to a specific set of n nodes, chosen out of m memory nodes. For example, for k hosts, m memory nodes and n-way replication, the total number of multicast groups needed can be k×mCn, where mCn is the number of combinations of n-tuples from m memory nodes. A memory node could subscribe to k×mC(n-1) of those multicast groups. This can enable each memory node to have a peer replica group with all the combinations of (n−1) of its peers. A host could subscribe to mCn groups.
A source PMA 1700 can maintain a running sequence number for each of the multicast replica groups assigned to it. The sequence number can be included in a packet transmitted to switch 1725 and incremented for the next packet transmission. When the multicast packet arrives at the receiver (e.g., memory node PMA 1750), memory node PMA 1750 can send an acknowledgement (Ack) with the received sequence number to the sender PMA 1700. The Acks can be sent by the memory node PMA 1750 as unicast packets to the source PMA's network address. When the source PMA 1700 receives all the Acks from the n memory nodes of the multicast group with transmitted sequence numbers, the write operation can be considered complete.
Determination if a re-transmit of a write operation to a particular memory node is to take place can be based on receipt of all Acks from each memory node's PMA within a multicast group within a timed period. A timer can be used to determine if all Acks from the n memory nodes in a multicast group were received within a time from original write is sent by a source PMA 1700. In some examples, receipt of all the Acks from host nodes with transmitted sequence numbers can be tracked by the host PMA 1700. Upon expiration of the timer, the host PMA 1700 can re-transmit the data to the memory nodes from which Acks were not received. This re-transmit can be performed as unicast packets to avoid sending duplicates to the nodes that have already received it.
In some examples, switch 1725 can perform operations of source PMA 1700 described herein to manage re-transmissions instead of source PMA 1700 so that nodes send Acks to switch 1725 and switch 1725 can send the packet just to the destination PMA from which an Ack was not received within a timed period.
In some cases, use of a timer to determine whether to re-transmit a packet can add latency to packet loss detection and re-transmission as packet loss occurs after a timer expiration and not before timer expiration. Some embodiments can reduce latency by providing that memory nodes send Acks and identify a write operation to the same multicast group members instead of unicast to the host PMA. In other words, memory nodes can send Acks to all the members of its multicast group including the peer receivers. A PMA 1750 can be configured with identifiers of memory nodes in a multicast group and hence maintain a scoreboard of Acks for a write operation from the memory nodes belonging to that replica group. If a receiver of a memory node does not receive the original packet but receives the Ack, the receiver can signal to source PMA 1700 that a packet was lost by sending a negative acknowledgement (Nack) to the sender PMA 1700 (e.g., via unicast). Sender PMA 1700 can detect lost packets and re-transmit lost packets to one or more destination PMAs 1750 instead of waiting for a timer to expire. In some examples, if a receiver of a memory node does not receive the original packet but receives an Ack from a memory node in its multicast group, the receiver can signal to one or more memory nodes in its multicast group that provided an Ack by sending a Nack to the one or more memory nodes in its multicast group and a memory node in its multicast group can send the packet to the PMA 1750 and perform the packet re-transmission.
If a memory node repeatedly fails to generate Acks, the switch can inform the host of repeated failure and the host or other entity (e.g., orchestrator) can change a memory group to exclude the failing node or generate a report to an administrator to identify the failing node.
At 1804, the source memory interface device can determine whether to transmit a packet to a memory node or a group of memory nodes. For example, the source memory interface device can determine to transmit the packet to a group of memory nodes if a destination address of the packet is associated with a destination memory node that is part of a memory node group. At least in such case, the packet can be identified as to be sent to multiple memory nodes and the process can continue to 1806. For example, the source memory interface device can determine to transmit the packet to a single memory node or multiple memory nodes if a destination address of the packet is not associated with a destination memory node that is part of a memory node group. In such case, the packet can be identified as to be sent to a single or multiple memory nodes and the process can continue to 1820.
At 1806, the source memory interface device can transmit the packet to a switch and include a multicast identifier in the packet. The switch can be configured to identify network addresses of destination memory nodes to which to send the packet or replicas thereof based on the multicast identifier. For example, the switch can be configured with a table of destination IP addresses corresponding to a multicast identifier.
At 1808, the source memory interface device can determine whether to re-transmit any packet. For example, if the source memory interface device receives acknowledgements of receipt from the memory nodes in the memory node group within a time window, the source memory interface device can determine that no packet re-transmission is needed (not shown in
At 1810, the packet can be re-transmitted to one or more memory nodes in the memory node group identified as not having received the packet. For example, the source memory interface device can identify any packet that was not received by a unique sequence identifier or other identifier and the source memory interface device can re-transmit such packet to one or more memory nodes in the memory node group identified as not having received the packet. In some examples, the switch can perform 1810 and re-transmit a particular packet.
In some examples, memory nodes can send acknowledgements to other memory nodes in the memory group, and a memory node that does not receive a packet identified as acknowledged, within a time window after receipt of a first acknowledgement from another memory node in the memory group, can request the source memory interface device or the switch to re-transmit such packet. The packet that was not received can be identified by a unique sequence identifier or other identifier.
Network interface 1900 can include transceiver 1902, processors 1904, transmit queue 1906, receive queue 1908, memory 1910, and bus interface 1912, and DMA engine 1952. Transceiver 1902 can be capable of receiving and transmitting packets in conformance with the applicable protocols such as Ethernet as described in IEEE 802.3, although other protocols may be used. Transceiver 1902 can receive and transmit packets from and to a network via a network medium (not depicted). Transceiver 1902 can include PHY circuitry 1914 and media access control (MAC) circuitry 1916. PHY circuitry 1914 can include encoding and decoding circuitry (not shown) to encode and decode data packets according to applicable physical layer specifications or standards. MAC circuitry 1916 can be configured to perform MAC address filtering on received packets, process MAC headers of received packets by verifying data integrity, remove preambles and padding, and provide packet content for processing by higher layers. MAC circuitry 1916 can be configured to assemble data to be transmitted into packets, that include destination and source addresses along with network control information and error detection hash values.
Processors 1904 can be any a combination of a: processor, core, graphics processing unit (GPU), field programmable gate array (FPGA), application specific integrated circuit (ASIC), or other programmable hardware device that allow programming of network interface 1900. For example, a “smart network interface” or SmartNIC can provide packet processing capabilities in the network interface using processors 1904. In some examples, processors 1904 can be implemented as a processor component for a SmartNIC.
Packet allocator 1924 can provide distribution of received packets for processing by multiple CPUs or cores using timeslot allocation described herein or RSS. When packet allocator 1924 uses RSS, packet allocator 1924 can calculate a hash or make another determination based on contents of a received packet to determine which CPU or core is to process a packet.
Interrupt coalesce 1922 can perform interrupt moderation whereby network interface interrupt coalesce 1922 waits for multiple packets to arrive, or for a time-out to expire, before generating an interrupt to host system to process received packet(s). Receive Segment Coalescing (RSC) can be performed by network interface 1900 whereby portions of incoming packets are combined into segments of a packet. Network interface 1900 provides this coalesced packet to an application.
Direct memory access (DMA) engine 1952 can copy a packet header, packet payload, and/or descriptor directly from host memory to the network interface or vice versa, instead of copying the packet to an intermediate buffer at the host and then using another copy operation from the intermediate buffer to the destination buffer. In some embodiments, multiple DMA engines are available for transfer of contents of packets to a destination memory associated with a host device or a destination memory associated with an accelerator device.
Memory 1910 can be any type of volatile or non-volatile memory device and can store any queue or instructions used to program network interface 1900. Transmit queue 1906 can include data or references to data for transmission by network interface. Receive queue 1908 can include data or references to data that was received by network interface from a network. Descriptor queues 1920 can include descriptors that reference data or packets in transmit queue 1906 or receive queue 1908 and corresponding destination memory regions. Bus interface 1912 can provide an interface with host device (not depicted). For example, bus interface 1912 can be compatible with PCI, PCI Express, PCI-x, Serial ATA, and/or USB compatible interface (although other interconnection standards may be used).
In one example, system 2000 includes interface 2012 coupled to processor 2010, which can represent a higher speed interface or a high throughput interface for system components that needs higher bandwidth connections, such as memory subsystem 2020, graphics interface components 2040, or accelerators 2042. Interface 2012 represents an interface circuit, which can be a standalone component or integrated onto a processor die. Where present, graphics interface 2040 interfaces to graphics components for providing a visual display to a user of system 2000. In one example, graphics interface 2040 can drive a high definition (HD) display that provides an output to a user. High definition can refer to a display having a pixel density of approximately 100 PPI (pixels per inch) or greater and can include formats such as full HD (e.g., 1080p), retina displays, 4K (ultra-high definition or UHD), or others. In one example, the display can include a touchscreen display. In one example, graphics interface 2040 generates a display based on data stored in memory 2030 or based on operations executed by processor 2010 or both. In one example, graphics interface 2040 generates a display based on data stored in memory 2030 or based on operations executed by processor 2010 or both.
Accelerators 2042 can be a programmable or fixed function offload engine that can be accessed or used by a processor 2010. For example, an accelerator among accelerators 2042 can provide compression (DC) capability, cryptography services such as public key encryption (PKE), cipher, hash/authentication capabilities, decryption, or other capabilities or services. In some embodiments, in addition or alternatively, an accelerator among accelerators 2042 provides field select controller capabilities as described herein. In some cases, accelerators 2042 can be integrated into a CPU socket (e.g., a connector to a motherboard or circuit board that includes a CPU and provides an electrical interface with the CPU). For example, accelerators 2042 can include a single or multi-core processor, graphics processing unit, logical execution unit single or multi-level cache, functional units usable to independently execute programs or threads, application specific integrated circuits (ASICs), neural network processors (NNPs), programmable control logic, and programmable processing elements such as field programmable gate arrays (FPGAs). Accelerators 2042 can provide multiple neural networks, processor cores, or graphics processing units can be made available for use by artificial intelligence (AI) or machine learning (ML) models. For example, the AI model can use or include any or a combination of: a reinforcement learning scheme, Q-learning scheme, deep-Q learning, or Asynchronous Advantage Actor-Critic (A3C), combinatorial neural network, recurrent combinatorial neural network, or other AI or ML model. Multiple neural networks, processor cores, or graphics processing units can be made available for use by AI or ML models.
Memory subsystem 2020 represents the main memory of system 2000 and provides storage for code to be executed by processor 2010, or data values to be used in executing a routine. Memory subsystem 2020 can include one or more memory devices 2030 such as read-only memory (ROM), flash memory, one or more varieties of random access memory (RAM) such as DRAM, or other memory devices, or a combination of such devices. Memory 2030 stores and hosts, among other things, operating system (OS) 2032 to provide a software platform for execution of instructions in system 2000. Additionally, applications 2034 can execute on the software platform of OS 2032 from memory 2030. Applications 2034 represent programs that have their own operational logic to perform execution of one or more functions. Processes 2036 represent agents or routines that provide auxiliary functions to OS 2032 or one or more applications 2034 or a combination. OS 2032, applications 2034, and processes 2036 provide software logic to provide functions for system 2000. In one example, memory subsystem 2020 includes memory controller 2022, which is a memory controller to generate and issue commands to memory 2030. It will be understood that memory controller 2022 could be a physical part of processor 2010 or a physical part of interface 2012. For example, memory controller 2022 can be an integrated memory controller, integrated onto a circuit with processor 2010.
In some examples, OS 2032 can determine a capability of a device associated with a device driver. For example, OS 2032 can receive an indication of a capability of a device (e.g., NIC 2050 or a storage configuration interface) to transmit packets to members of a multicast group or re-transmit packets in accordance with embodiments described herein. OS 2032 can request a driver to enable or disable NIC 2050 to perform any of the capabilities described herein. In some examples, OS 2032, itself, can enable or disable NIC 2050 to perform any of the capabilities described herein. OS 2032 can provide requests (e.g., from an application or VM) to NIC 2050 to utilize one or more capabilities of NIC 2050. For example, any application can request use or non-use of any of capabilities described herein by NIC 2050.
While not specifically illustrated, it will be understood that system 2000 can include one or more buses or bus systems between devices, such as a memory bus, a graphics bus, interface buses, or others. Buses or other signal lines can communicatively or electrically couple components together, or both communicatively and electrically couple the components. Buses can include physical communication lines, point-to-point connections, bridges, adapters, controllers, or other circuitry or a combination. Buses can include, for example, one or more of a system bus, a Peripheral Component Interconnect (PCI) bus, a HyperTransport or industry standard architecture (ISA) bus, a small computer system interface (SCSI) bus, a universal serial bus (USB), or an Institute of Electrical and Electronics Engineers (IEEE) standard 1394 bus.
In one example, system 2000 includes interface 2014, which can be coupled to interface 2012. In one example, interface 2014 represents an interface circuit, which can include standalone components and integrated circuitry. In one example, multiple user interface components or peripheral components, or both, couple to interface 2014. Network interface 2050 provides system 2000 the ability to communicate with remote devices (e.g., servers or other computing devices) over one or more networks. Network interface 2050 can include an Ethernet adapter, wireless interconnection components, cellular network interconnection components, USB (universal serial bus), or other wired or wireless standards-based or proprietary interfaces. Network interface 2050 can transmit data to a remote device, which can include sending data stored in memory. Network interface 2050 can receive data from a remote device, which can include storing received data into memory. Various embodiments can be used in connection with network interface 2050, processor 2010, and memory subsystem 2020.
In one example, system 2000 includes one or more input/output (I/O) interface(s) 2060. I/O interface 2060 can include one or more interface components through which a user interacts with system 2000 (e.g., audio, alphanumeric, tactile/touch, or other interfacing). Peripheral interface 2070 can include any hardware interface not specifically mentioned above. Peripherals refer generally to devices that connect dependently to system 2000. A dependent connection is one where system 2000 provides the software platform or hardware platform or both on which operation executes, and with which a user interacts.
In one example, system 2000 includes storage subsystem 2080 to store data in a nonvolatile manner. In one example, in certain system implementations, at least certain components of storage 2080 can overlap with components of memory subsystem 2020. Storage subsystem 2080 includes storage device(s) 2084, which can be or include any conventional medium for storing large amounts of data in a nonvolatile manner, such as one or more magnetic, solid state, or optical based disks, or a combination. Storage 2084 holds code or instructions and data 2086 in a persistent state (e.g., the value is retained despite interruption of power to system 2000). Storage 2084 can be generically considered to be a “memory,” although memory 2030 is typically the executing or operating memory to provide instructions to processor 2010. Whereas storage 2084 is nonvolatile, memory 2030 can include volatile memory (e.g., the value or state of the data is indeterminate if power is interrupted to system 2000). In one example, storage subsystem 2080 includes controller 2082 to interface with storage 2084. In one example controller 2082 is a physical part of interface 2014 or processor 2010 or can include circuits or logic in both processor 2010 and interface 2014.
A volatile memory is memory whose state (and therefore the data stored in it) is indeterminate if power is interrupted to the device. Dynamic volatile memory requires refreshing the data stored in the device to maintain state. One example of dynamic volatile memory incudes DRAM (Dynamic Random Access Memory), or some variant such as Synchronous DRAM (SDRAM). Another example of volatile memory includes cache or static random access memory (SRAM). A memory subsystem as described herein may be compatible with a number of memory technologies, such as DDR3 (Double Data Rate version 3, original release by JEDEC (Joint Electronic Device Engineering Council) on Jun. 27, 2007). DDR4 (DDR version 4, initial specification published in September 2012 by JEDEC), DDR4E (DDR version 4), LPDDR3 (Low Power DDR version3, JESD209-3B, August 2013 by JEDEC), LPDDR4) LPDDR version 4, JESD209-4, originally published by JEDEC in August 2014), WIO2 (Wide Input/output version 2, JESD229-2 originally published by JEDEC in August 2014, HBM (High Bandwidth Memory, JESD325, originally published by JEDEC in October 2013, LPDDR5 (currently in discussion by JEDEC), HBM2 (HBM version 2), currently in discussion by JEDEC, or others or combinations of memory technologies, and technologies based on derivatives or extensions of such specifications. The JEDEC standards are available at www.jedec.org.
A non-volatile memory (NVM) device is a memory whose state is determinate even if power is interrupted to the device. In one embodiment, the NVM device can comprise a block addressable memory device, such as NAND technologies, or more specifically, multi-threshold level NAND flash memory (for example, Single-Level Cell (“SLC”), Multi-Level Cell (“MLC”), Quad-Level Cell (“QLC”), Tri-Level Cell (“TLC”), or some other NAND). A NVM device can also comprise a byte-addressable write-in-place three dimensional cross point memory device, or other byte addressable write-in-place NVM device (also referred to as persistent memory), such as single or multi-level Phase Change Memory (PCM) or phase change memory with a switch (PCMS), Intel® Optane™ memory, NVM devices that use chalcogenide phase change material (for example, chalcogenide glass), resistive memory including metal oxide base, oxygen vacancy base and Conductive Bridge Random Access Memory (CB-RAM), nanowire memory, ferroelectric random access memory (FeRAM, FRAM), magneto resistive random access memory (MRAM) that incorporates memristor technology, spin transfer torque (STT)-MRAM, a spintronic magnetic junction memory based device, a magnetic tunneling junction (MTJ) based device, a DW (Domain Wall) and SOT (Spin Orbit Transfer) based device, a thyristor based memory device, or a combination of any of the above, or other memory.
A power source (not depicted) provides power to the components of system 2000. More specifically, power source typically interfaces to one or multiple power supplies in system 2000 to provide power to the components of system 2000. In one example, the power supply includes an AC to DC (alternating current to direct current) adapter to plug into a wall outlet. Such AC power can be renewable energy (e.g., solar power) power source. In one example, power source includes a DC power source, such as an external AC to DC converter. In one example, power source or power supply includes wireless charging hardware to charge via proximity to a charging field. In one example, power source can include an internal battery, alternating current supply, motion-based power supply, solar power supply, or fuel cell source.
In an example, system 2000 can be implemented using interconnected compute sleds of processors, memories, storages, network interfaces, and other components. High speed connections can be used such as: Ethernet (IEEE 802.3), remote direct memory access (RDMA), InfiniBand, Internet Wide Area RDMA Protocol (iWARP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), quick UDP Internet Connections (QUIC), RDMA over Converged Ethernet (RoCE), Peripheral Component Interconnect express (PCIe), Intel® QuickPath Interconnect (QPI), Intel® Ultra Path Interconnect (UPI), Intel® On-Chip System Fabric (IOSF), Omnipath, Compute Express Link (CXL), HyperTransport, high-speed fabric, NVLink, Advanced Microcontroller Bus Architecture (AMB A) interconnect, OpenCAPI, Gen-Z, Cache Coherent Interconnect for Accelerators (CCIX), 3GPP Long Term Evolution (LTE) (4G), 3GPP 5G, DisplayPort, embedded DisplayPort, MIPI, HDMI, Infinity Fabric (IF), and successors or variations thereof.
Embodiments herein may be implemented in various types of computing and networking equipment, such as switches, routers, racks, and blade servers such as those employed in a data center and/or server farm environment. The servers used in data centers and server farms comprise arrayed server configurations such as rack-based servers or blade servers. These servers are interconnected in communication via various network provisions, such as partitioning sets of servers into Local Area Networks (LANs) with appropriate switching and routing facilities between the LANs to form a private Intranet. For example, cloud hosting facilities may typically employ large data centers with a multitude of servers. A blade comprises a separate computing platform that is configured to perform server-type functions, that is, a “server on a card.” Accordingly, each blade includes components common to conventional servers, including a main printed circuit board (main board) providing internal wiring (e.g., buses) for coupling appropriate integrated circuits (ICs) and other components mounted to the board.
In some examples, network interface and other embodiments described herein can be used in connection with a base station (e.g., 3G, 4G, 5G and so forth), macro base station (e.g., 5G networks), picostation (e.g., an IEEE 802.11 compatible access point), nanostation (e.g., for Point-to-MultiPoint (PtMP) applications), on-premises data centers, off-premises data centers, edge network elements, fog network elements, and/or hybrid data centers (e.g., data center that use virtualization, cloud and software-defined networking to deliver application workloads across physical data centers and distributed multi-cloud environments).
Various examples may be implemented using hardware elements, software elements, or a combination of both. In some examples, hardware elements may include devices, components, processors, microprocessors, circuits, circuit elements (e.g., transistors, resistors, capacitors, inductors, and so forth), integrated circuits, ASICs, PLDs, DSPs, FPGAs, memory units, logic gates, registers, semiconductor device, chips, microchips, chip sets, and so forth. In some examples, software elements may include software components, programs, applications, computer programs, application programs, system programs, machine programs, operating system software, middleware, firmware, software modules, routines, subroutines, functions, methods, procedures, software interfaces, APIs, instruction sets, computing code, computer code, code segments, computer code segments, words, values, symbols, or any combination thereof. Determining whether an example is implemented using hardware elements and/or software elements may vary in accordance with any number of factors, such as desired computational rate, power levels, heat tolerances, processing cycle budget, input data rates, output data rates, memory resources, data bus speeds and other design or performance constraints, as desired for a given implementation. A processor can be one or more combination of a hardware state machine, digital control logic, central processing unit, or any hardware, firmware and/or software elements.
Some examples may be implemented using or as an article of manufacture or at least one computer-readable medium. A computer-readable medium may include a non-transitory storage medium to store logic. In some examples, the non-transitory storage medium may include one or more types of computer-readable storage media capable of storing electronic data, including volatile memory or non-volatile memory, removable or non-removable memory, erasable or non-erasable memory, writeable or re-writeable memory, and so forth. In some examples, the logic may include various software elements, such as software components, programs, applications, computer programs, application programs, system programs, machine programs, operating system software, middleware, firmware, software modules, routines, subroutines, functions, methods, procedures, software interfaces, API, instruction sets, computing code, computer code, code segments, computer code segments, words, values, symbols, or any combination thereof.
According to some examples, a computer-readable medium may include a non-transitory storage medium to store or maintain instructions that when executed by a machine, computing device or system, cause the machine, computing device or system to perform methods and/or operations in accordance with the described examples. The instructions may include any suitable type of code, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, and the like. The instructions may be implemented according to a predefined computer language, manner or syntax, for instructing a machine, computing device or system to perform a certain function. The instructions may be implemented using any suitable high-level, low-level, object-oriented, visual, compiled and/or interpreted programming language.
One or more aspects of at least one example may be implemented by representative instructions stored on at least one machine-readable medium which represents various logic within the processor, which when read by a machine, computing device or system causes the machine, computing device or system to fabricate logic to perform the techniques described herein. Such representations, known as “IP cores” may be stored on a tangible, machine readable medium and supplied to various customers or manufacturing facilities to load into the fabrication machines that actually make the logic or processor.
The appearances of the phrase “one example” or “an example” are not necessarily all referring to the same example or embodiment. Any aspect described herein can be combined with any other aspect or similar aspect described herein, regardless of whether the aspects are described with respect to the same figure or element. Division, omission or inclusion of block functions depicted in the accompanying figures does not infer that the hardware components, circuits, software and/or elements for implementing these functions would necessarily be divided, omitted, or included in embodiments.
Some examples may be described using the expression “coupled” and “connected” along with their derivatives. These terms are not necessarily intended as synonyms for each other. For example, descriptions using the terms “connected” and/or “coupled” may indicate that two or more elements are in direct physical or electrical contact with each other. The term “coupled,” however, may also mean that two or more elements are not in direct contact with each other, but yet still co-operate or interact with each other.
The terms “first,” “second,” and the like, herein do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items. The term “asserted” used herein with reference to a signal denote a state of the signal, in which the signal is active, and which can be achieved by applying any logic level either logic 0 or logic 1 to the signal. The terms “follow” or “after” can refer to immediately following or following after some other event or events. Other sequences of steps may also be performed according to alternative embodiments. Furthermore, additional steps may be added or removed depending on the particular applications. Any combination of changes can be used and one of ordinary skill in the art with the benefit of this disclosure would understand the many variations, modifications, and alternative embodiments thereof.
Disjunctive language such as the phrase “at least one of X, Y, or Z,” unless specifically stated otherwise, is otherwise understood within the context as used in general to present that an item, term, etc., may be either X, Y, or Z, or any combination thereof (e.g., X, Y, and/or Z). Thus, such disjunctive language is not generally intended to, and should not, imply that certain embodiments require at least one of X, at least one of Y, or at least one of Z to each be present. Additionally, conjunctive language such as the phrase “at least one of X, Y, and Z,” unless specifically stated otherwise, should also be understood to mean X, Y, Z, or any combination thereof, including “X, Y, and/or Z.”′
Example 1 includes any example and includes a method comprising: designating a packet for transmission to multiple memory nodes by including a multicast identifier of a memory node group and transmitting the packet to a destination network device, wherein the multicast identifier of the memory node group in the packet is to cause at least one intermediate network device to multicast the packet to multiple memory nodes.
Example 2 includes any example, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
Example 3 includes any example, wherein the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
Example 4 includes any example, and includes re-transmitting a packet to a memory node in the memory node group based on receipt of a negative acknowledgement for the packet or failure to receive an acknowledgement of receipt of the packet within a time window.
Example 5 includes any example, wherein a host memory node interface performs the re-transmitting.
Example 6 includes any example, wherein the intermediate network device performs the re-transmitting.
Example 7 includes any example, and includes receiving a request to re-transmit a packet from a memory node that does not receive a packet associated with receipt acknowledgement sent to the memory node from other memory nodes in the memory node group and re-transmitting the requested re-transmit packet.
Example 8 includes any example, and includes a computer-readable medium comprising instructions stored thereon, that if executed by one or more processors, cause the one or more processors to: configure a network interface to: generate a packet for transmission to multiple memory nodes by inclusion of a multicast identifier of a memory node group in the packet and transmit the packet to a destination network device, wherein the multicast identifier of the memory node group in the packet is to cause an intermediate network device to multicast the packet to multiple memory nodes.
Example 9 includes any example, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
Example 10 includes any example, wherein the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
Example 11 includes any example, comprising instructions stored thereon, that if executed by one or more processors, cause the one or more processors to: configure the network interface to: in response to a request for a packet re-transmission, re-transmit a packet from a memory node that does not receive a packet associated with receipt acknowledgement sent to the memory node from other memory nodes in the memory node group.
Example 12 includes any example, and includes an apparatus comprising: a network interface configured to: access data designated for transmission in at least one packet to multiple memory nodes by inclusion of an multicast identifier of a memory node group and transmit the at least one packet to a destination network device, wherein the multicast identifier of the memory node group in the at least one packet is to cause an intermediate network device to multicast the packet to multiple memory nodes.
Example 13 includes any example, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
Example 14 includes any example, wherein the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
Example 15 includes any example, wherein the network interface is to: re-transmit a packet to a memory node in the memory node group based on receipt of a negative acknowledgement for the packet or failure to receive an acknowledgement of receipt of the packet within a time window.
Example 16 includes any example, wherein the network interface is to: in response to a request for a packet re-transmission, re-transmit a packet from a memory node that does not receive a packet associated with receipt acknowledgement sent to the memory node from other memory nodes in the memory node group.
Example 17 includes any example, comprising one or more of a server, rack, or data center coupled to the network interface, wherein the one or more of a server, rack, or data center is to provide the data.
Example 18 includes any example, and includes an apparatus comprising: a network element, when operational, to: receive a packet and based on the received packet including a multicast identifier of a memory node group, multicasting the received packet to multiple memory nodes of the memory node group.
Example 19 includes any example, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
Example 20 includes any example, wherein the network element comprises a switch, network interface, or a router.
Example 21 includes any example, wherein the network element, when operational, is to: re-transmit a packet to a memory node in the memory node group based on receipt of a negative acknowledgement for the packet or failure to receive an acknowledgement of receipt of the packet within a time window.
Claims
1. A method comprising:
- designating a packet for transmission to multiple memory nodes by including a multicast identifier of a memory node group and
- transmitting the packet to a destination network device, wherein the multicast identifier of the memory node group in the packet is to cause at least one intermediate network device to multicast the packet to multiple memory nodes.
2. The method of claim 1, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
3. The method of claim 1, wherein the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
4. The method of claim 1, comprising:
- re-transmitting a packet to a memory node in the memory node group based on receipt of a negative acknowledgement for the packet or failure to receive an acknowledgement of receipt of the packet within a time window.
5. The method of claim 4, wherein a host memory node interface performs the re-transmitting.
6. The method of claim 4, wherein the intermediate network device performs the re-transmitting.
7. The method of claim 1, comprising:
- receiving a request to re-transmit a packet from a memory node that does not receive a packet associated with receipt acknowledgement sent to the memory node from other memory nodes in the memory node group and
- re-transmitting the requested re-transmit packet.
8. A computer-readable medium comprising instructions stored thereon, that if executed by one or more processors, cause the one or more processors to:
- configure a network interface to:
- generate a packet for transmission to multiple memory nodes by inclusion of a multicast identifier of a memory node group in the packet and transmit the packet to a destination network device, wherein the multicast identifier of the memory node group in the packet is to cause an intermediate network device to multicast the packet to multiple memory nodes.
9. The computer-readable medium of claim 8, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
10. The computer-readable medium of claim 8, wherein the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
11. The computer-readable medium of claim 8, comprising instructions stored thereon, that if executed by one or more processors, cause the one or more processors to:
- configure the network interface to:
- in response to a request for a packet re-transmission, re-transmit a packet from a memory node that does not receive a packet associated with receipt acknowledgement sent to the memory node from other memory nodes in the memory node group.
12. An apparatus comprising:
- a network interface configured to:
- access data designated for transmission in at least one packet to multiple memory nodes by inclusion of an multicast identifier of a memory node group and
- transmit the at least one packet to a destination network device, wherein the multicast identifier of the memory node group in the at least one packet is to cause an intermediate network device to multicast the packet to multiple memory nodes.
13. The apparatus of claim 12, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
14. The apparatus of claim 12, wherein the intermediate network device comprises a switch configured to determine network addresses of memory nodes associated with the multicast identifier of the memory node group.
15. The apparatus of claim 12, wherein the network interface is to:
- re-transmit a packet to a memory node in the memory node group based on receipt of a negative acknowledgement for the packet or failure to receive an acknowledgement of receipt of the packet within a time window.
16. The apparatus of claim 12, wherein the network interface is to:
- in response to a request for a packet re-transmission, re-transmit a packet from a memory node that does not receive a packet associated with receipt acknowledgement sent to the memory node from other memory nodes in the memory node group.
17. The apparatus of claim 12, comprising one or more of a server, rack, or data center coupled to the network interface, wherein the one or more of a server, rack, or data center is to provide the data.
18. An apparatus comprising:
- a network element, when operational, to: receive a packet and based on the received packet including a multicast identifier of a memory node group, multicasting the received packet to multiple memory nodes of the memory node group.
19. The apparatus of claim 18, wherein a memory node comprises a memory pool that includes one or more of: volatile memory, non-volatile memory, or persistent memory.
20. The apparatus of claim 18, wherein the network element comprises a switch, network interface, or a router.
21. The apparatus of claim 18, wherein the network element, when operational, is to:
- re-transmit a packet to a memory node in the memory node group based on receipt of a negative acknowledgement for the packet or failure to receive an acknowledgement of receipt of the packet within a time window.
Type: Application
Filed: Nov 24, 2020
Publication Date: Mar 11, 2021
Inventors: Sujoy SEN (Beaverton, OR), Thomas E. WILLIS (Redwood City, CA), Durgesh SRIVASTAVA (Cupertino, CA), Marcelo CINTRA (Braunschweig), Bassam N. COURY (Portland, OR)
Application Number: 17/103,674