Abstract: Transparent memory management for over-subscribed accelerators is disclosed. A request from a remote initiator to execute a workload on a shared accelerator is received at a host system comprising the shared accelerator. A determination is made that there is insufficient physical memory of the accelerator to accommodate the request from the remote initiator. Responsive to determining that there is insufficient physical memory of the accelerator. An allocation of host system memory is requested for the remote initiator from the host system. A mapping between the remote initiator and the allocation of host system memory is then created.

Abstract: Managing server nodes on a management network with modified service discovery protocol messages including determining that a first management server requires management data for installation on a server node, wherein the first management server and the server node are communicatively coupled to a management network; determining that the required management data for installation on the server node is not stored on the first management server; broadcasting, on the management network, a first modified service discovery protocol (SDP) message requesting the required management data for installation on the server node; receiving, from a second management server communicatively coupled to the management network, a second modified SDP message providing the required management data for installation on the server node; and installing the required management data on the server node.

Abstract: A CPU package includes an encryption and decryption module disposed in a communication path between an instruction path of a processor core and a data register that is externally accessible through a debug port, and a key store accessible to the module. The module is configured to encrypt and store data in the data register for each of a plurality of processes being handled in the instruction path, wherein data owned by each process is encrypted and decrypted by the module using an encryption key assigned to the process. The key store is configured to store the encryption key assigned to each of a plurality of processes, wherein the key store is inaccessible outside the CPU package. The data is only decrypted for a requesting process having a process identifier that matches the process identifier stored in the processor data structure along with the requested data.

Abstract: A method includes migrating a workload from a first server to a second server, wherein the first and second servers are within a management network. The method further includes transferring one or more configuration parameter(s) from the first server to the second server, wherein the one or more parameter(s) enables continued performance of the workload on the second server.

Abstract: For system management applied to a computer system, a power supply of the computer system starts to power a motherboard and a CPU thereon. A reset holding module in a system management controller holds the CPU in a Power-on Reset (PoR) state. The system management controller executes an operation requested by a user. The reset holding module releases the CPU from the PoR state in response to the system management controller completing the operation.

Abstract: Managing server nodes on a management network with modified service discovery protocol messages including determining that a first management server requires management data for installation on a server node, wherein the first management server and the server node are communicatively coupled to a management network; determining that the required management data for installation on the server node is not stored on the first management server; broadcasting, on the management network, a first modified service discovery protocol (SDP) message requesting the required management data for installation on the server node; receiving, from a second management server communicatively coupled to the management network, a second modified SDP message providing the required management data for installation on the server node; and installing the required management data on the server node.

Abstract: An apparatus includes one or more processor core, trusted key store, memory controller, and a memory module. The memory controller includes an encryption/decryption module that encrypts data being stored to the memory module for a guest OS being executed by the processor core(s) and that decrypts data being read from the memory module for the guest OS. Data owned by the guest OS is encrypted and decrypted by the encryption/decryption module using an encryption key stored by the trusted key store in association with the guest OS. A method encrypts data owned by the guest OS using the encryption key assigned to the guest OS and stores the encrypted data on a memory module, wherein the encrypted data is stored in association with the process identifier of the guest OS, and decrypts the encrypted data using the guest OS encryption key and provides the decrypted data to the guest OS.

Abstract: Managing thermal cycles of air temperature within a server rack includes: monitoring air temperature within the server rack; determining that the monitored temperature has fallen below a predetermined minimum threshold; and increasing air temperature within the server rack including capturing warm ambient air.

Abstract: Topology discovery between compute nodes and interconnect switches including creating, on an interconnect switch, a virtual topology discovery device for a first port, wherein the interconnect switch is coupled to a compute node via the first port, and wherein the virtual topology discovery device comprises a port identifier for the first port; mapping the virtual topology discovery device to the first port; receiving an inventory request from the compute node via the first port; routing the inventory request to the virtual topology discovery device for the first port; and sending, from the virtual topology discovery device for the first port, the port identifier to the compute node.

Abstract: Performing maintenance tasks on composed systems during workload execution including monitoring a performance of a compute element during the execution of a workload, wherein the compute element is mapped to a composed system executing the workload, and wherein the compute element and the composed system are within a pod of composable compute elements; determining, based on the performance of the compute element, that the compute element has a pending maintenance task; unmapping, from the composed system during the execution of the workload, the compute element with the pending maintenance task; performing the maintenance task on the unmapped compute element during the execution of the workload by the composed system; and remapping the compute element to the composed system during the execution of the workload.

Abstract: Managing composable compute systems with support for hyperconverged software defined storage includes: monitoring a composable pod of computing elements, wherein the composable pod of computing elements comprises a first composed server and a first data drive, wherein the first composed server comprises at least one of the computing elements of the composable pod, and wherein the first data drive is configured for attachment to a composable system executing hyperconverged storage software; detecting that a workload on the first composed server of the composable pod targets data on the first data drive of the composable pod; determining that the first data drive of the composable pod is not mapped to the first composed server; and mapping the first data drive containing the data targeted by the workload to the first composed server hosting the workload.

Abstract: A method includes providing, by a network switch, a production VLAN that supports communication between a plurality of servers connected to the switch; and detecting a new physical connection of a cable between a first server and a first switch port of the switch. The method further includes automatically allocating, by the switch, a first provisioning VLAN to the first server, wherein communications coming from the first server are directed to the first provisioning VLAN, wherein the first provisioning VLAN includes a provisioning server, and wherein the first provisioning VLAN is isolated against communication with the production VLAN. An apparatus that performs the method is also provided.

Abstract: A method includes tracking, for each of a plurality of workloads, a wear rate incurred by a component of a given component type as a result of processing an instance of the workload on a compute node that includes the component. The method further comprises determining a remaining lifespan for a plurality of components in a plurality of compute nodes, and identifying a component from among the plurality of components having a remaining lifespan of less than a first threshold amount. Sill further, the method comprises identifying an instance of one of the workloads that needs processing and has a wear rate greater than a second threshold amount, and processing the identified workload instance on the identified component.

Abstract: Managing thermal cycles of air temperature within a server rack includes: monitoring air temperature within the server rack; determining that the monitored temperature has fallen below a predetermined minimum threshold; and increasing air temperature within the server rack including capturing warm ambient air.

Abstract: An apparatus, such as a computing node, includes a storage device for storing program instructions and a service processor for processing the program instructions to: send an out-of-band request for access to an operator panel, wherein the out-of-band request is sent over a local network domain operatively coupling a plurality of computing nodes; receive a first out-of-band reply identifying an address and one or more capabilities of an operator panel accessible over the local network domain; and negotiate an active link to the operator panel over the local network domain. Another apparatus is similar but further includes an operator panel operatively coupled to a service processor, the service processor for processing the program instructions to: receive the out-of-band request over the local network domain; send the first out-of-band message; and negotiate an active link between the operator panel and a computing node.

Abstract: An apparatus includes one or more processor core, trusted key store, memory controller, and a memory module. The memory controller includes an encryption/decryption module that encrypts data being stored to the memory module for a guest OS being executed by the processor core(s) and that decrypts data being read from the memory module for the guest OS. Data owned by the guest OS is encrypted and decrypted by the encryption/decryption module using an encryption key stored by the trusted key store in association with the guest OS. A method encrypts data owned by the guest OS using the encryption key assigned to the guest OS and stores the encrypted data on a memory module, wherein the encrypted data is stored in association with the process identifier of the guest OS, and decrypts the encrypted data using the guest OS encryption key and provides the decrypted data to the guest OS.

Abstract: A method includes providing, by a network switch, a production VLAN that supports communication between a plurality of servers connected to the switch; and detecting a new physical connection of a cable between a first server and a first switch port of the switch. The method further includes automatically allocating, by the switch, a first provisioning VLAN to the first server, wherein communications coming from the first server are directed to the first provisioning VLAN, wherein the first provisioning VLAN includes a provisioning server, and wherein the first provisioning VLAN is isolated against communication with the production VLAN. An apparatus that performs the method is also provided.

Abstract: A method includes migrating a workload from a first server to a second server, wherein the first and second servers are within a management network. The method further includes transferring one or more configuration parameter(s) from the first server to the second server, wherein the one or more parameter(s) enables continued performance of the workload on the second server.

Abstract: For system management applied to a computer system, a power supply of the computer system starts to power a motherboard and a CPU thereon. A reset holding module in a system management controller holds the CPU in a Power-on Reset (PoR) state. The system management controller executes an operation requested by a user. The reset holding module releases the CPU from the PoR state in response to the system management controller completing the operation.

Abstract: For system management applied to a computer system, a power supply of the computer system starts to power a motherboard and a CPU thereon. A reset holding module in a system management controller holds the CPU in a Power-on Reset (PoR) state. The system management controller executes an operation requested by a user. The reset holding module releases the CPU from the PoR state in response to the system management controller completing the operation.