Abstract: Distributed I/O virtualization includes receiving, at a first physical node in a plurality of physical nodes, an indication of a request to transfer data from an I/O device on the first physical node to a set of guest physical addresses. An operating system is executing collectively across the plurality of physical nodes. It further includes writing data from the I/O device to one or more portions of physical memory local to the first physical node. It further includes mapping the set of guest physical addresses to the written one or more portions of physical memory local to the first physical node.

Abstract: Maintaining cache coherency in the presence of a network attached memory is disclosed. A computer system includes a plurality of physical nodes. An operating system is run collectively across the plurality of physical nodes. The physical nodes are configured to communicate with a network attached memory. Based at least in part on an operation to be performed with respect to page on a first physical node included in the plurality of physical nodes, the network attached memory is configured to receive a message. The network attached memory is configured to perform an action based at least in part on the received message.

Abstract: Distributed I/O virtualization includes receiving, at a first physical node in a plurality of physical nodes, an indication of a request to transfer data from an I/O device on the first physical node to a set of guest physical addresses. An operating system is executing collectively across the plurality of physical nodes. It further includes writing data from the I/O device to one or more portions of physical memory local to the first physical node. It further includes mapping the set of guest physical addresses to the written one or more portions of physical memory local to the first physical node.

Abstract: Maintaining cache coherency in the presence of a network attached memory is disclosed. A computer system includes a plurality of physical nodes. An operating system is run collectively across the plurality of physical nodes. The physical nodes are configured to communicate with a network attached memory. Based at least in part on an operation to be performed with respect to page on a first physical node included in the plurality of physical nodes, the network attached memory is configured to receive a message. The network attached memory is configured to perform an action based at least in part on the received message.

Abstract: Initializing a computing system using dormant pages includes marking a set of guest physical addresses as dormant. It further includes, for each node in a plurality of physical nodes, designating a set of real physical addresses for zeroing. An operating system is executing collectively across the physical nodes.

Abstract: Dynamic adaptive reconfiguration of a computing system includes receiving a request to remove a first node in a plurality of physical nodes. An operating system is executing collectively across the plurality of physical nodes, and an application is running on the operating system. It further includes in response to the request, and while the application is running, evacuating virtualized resources associated with the first node to one or more other nodes in the plurality of physical nodes. It further includes subsequent to the evacuation of the virtualized resources, removing the first node from the plurality of physical nodes.

Abstract: Initializing a computing system using dormant pages includes marking a set of guest physical addresses as dormant. It further includes, for each node in a plurality of physical nodes, designating a set of real physical addresses for zeroing. An operating system is executing collectively across the physical nodes.

Abstract: Selective resource migration is disclosed. A computer system includes physical memory and a plurality of physical processors. Each of the processors has one or more cores and each core instantiates one or more virtual processors that executes program code. Each core is configured to invoke a hyper-kernel on its hosting physical processor when the core cannot access a portion of the physical memory needed by the core. The hyper-kernel selectively moves the needed memory closer to a location accessible by the physical processor or remaps the virtual processor to another core.

Abstract: Handling frequently accessed pages is disclosed. An indication is received of a stalling event caused by a requested portion of memory being inaccessible. It is determined that the requested portion of memory is a frequently updated portion of memory. The stalling event is handled based at least in part on the determination that the requested portion of memory is a frequently updated portion of memory.

Abstract: Dynamic adaptive reconfiguration of a computing system includes receiving a request to remove a first node in a plurality of physical nodes. An operating system is executing collectively across the plurality of physical nodes, and an application is running on the operating system. It further includes in response to the request, and while the application is running, evacuating virtualized resources associated with the first node to one or more other nodes in the plurality of physical nodes. It further includes subsequent to the evacuation of the virtualized resources, removing the first node from the plurality of physical nodes.

Abstract: Handling frequently accessed pages is disclosed. An indication is received of a stalling event caused by a requested portion of memory being inaccessible. It is determined that the requested portion of memory is a frequently updated portion of memory. The stalling event is handled based at least in part on the determination that the requested portion of memory is a frequently updated portion of memory.

Abstract: Initializing a software-defined server having software-defined NUMA domains includes, when booting a virtual environment defined by a set of hyper-kernels running on a plurality of physically interconnected computing nodes, accessing information associated with a software-defined NUMA domain configuration. It further includes, based at least in part on the accessed information, assigning software-defined NUMA domains to computing nodes. It further includes assigning virtualized resources to the software-defined NUMA domains. Handling a stalling event involving software-defined NUMA domains includes receiving an indication that a core or hyperthread which instantiates a virtual processor cannot access a virtualized resource needed by the core or hyperthread.

Abstract: Entanglement of pages and threads is disclosed. An indication is received of a stalling event caused by a requested portion of memory being inaccessible. It is determined that the requested portion of memory is an entangled portion of memory that is entangled with a physical node in a plurality of physical nodes. A type of the entangled portion of memory is determined. The stalling event is handled based at least in part on the determined type of the entangled portion of memory.

Abstract: Entanglement of pages and threads is disclosed. An indication is received of a stalling event caused by a requested portion of memory being inaccessible. It is determined that the requested portion of memory is an entangled portion of memory that is entangled with a physical node in a plurality of physical nodes. A type of the entangled portion of memory is determined. The stalling event is handled based at least in part on the determined type of the entangled portion of memory.

Abstract: Associating working sets and threads is disclosed. An indication of a stalling event is received. In response to receiving the indication of the stalling event, a state of a processor associated with the stalling event is saved. At least one of an identifier of a guest thread running in the processor and a guest physical address referenced by the processor is obtained from the saved processor state.

Abstract: Resource migration negotiation is disclosed. A request is received, from a remote physical node in a plurality of physical nodes, for a resource. An operating system is run collectively across the plurality of physical nodes. The request includes information pertaining to a guest thread running on the remote physical node. Based at least in part on at least some of the information included in the request, it is determined whether to send the requested resource or reject the request. A response is provided based at least in part on the determination.

Abstract: Selective resource migration is disclosed. A computer system includes physical memory and a plurality of physical processors. Each of the processors has one or more cores and each core instantiates one or more virtual processors that executes program code. Each core is configured to invoke a hyper-kernel on its hosting physical processor when the core cannot access a portion of the physical memory needed by the core. The hyper-kernel selectively moves the needed memory closer to a location accessible by the physical processor or remaps the virtual processor to another core.

Abstract: Hierarchical stalling strategies are disclosed. An indication is received of a stalling event caused by a requested resource being inaccessible. In response to receiving the indication of the stalling event, a set of cost functions usable to determine how to handle the stalling event is selected based at least in part on a type of the stalling event. The stalling event is handled based at least in part on an evaluation of the set of cost functions selected based at least in part on the type of the stalling event.

Abstract: Dynamic scheduling is disclosed. A plurality of physical nodes is included in a computer system. Each node includes a plurality of processors. Each processor includes a plurality of hyperthreads. In response to receiving an indication of an event occurring, a search is performed for a queue in a set of queues on which to place a virtual processor that had been waiting on the event. Queues in the set of queues correspond to hyperthreads in a physical node in the plurality of physical nodes. The queues in the set of queues are visited according to a predetermined traversal order.

Abstract: Hierarchical dynamic scheduling is disclosed. A plurality of physical nodes is included in a computer system. Each node includes a plurality of processors. Each processor includes a plurality of hyperthreads. An abstraction of the nodes, processors, and hyperthreads forms a hierarchy. Upon receiving an indication that a hyperthread should be assigned, a dynamic search of the hierarchy is performed, beginning at the leaf level, for a process to assign to the hyperthread.