Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Multi-core computers may implement a resource management layer between the operating system and resource-management-enabled parallel runtime systems. The resource management components and runtime systems may collectively implement dynamic co-scheduling of hardware contexts when executing multiple parallel applications, using a spatial scheduling policy that grants high priority to one application per hardware context and a temporal scheduling policy for re-allocating unused hardware contexts. The runtime systems may receive resources on a varying number of hardware contexts as demands of the applications change over time, and the resource management components may co-ordinate to leave one runnable software thread for each hardware context. Periodic check-in operations may be used to determine (at times convenient to the applications) when hardware contexts should be re-allocated. Over-subscription of worker threads may reduce load imbalances between applications.

Abstract: Multi-core computers may implement a resource management layer between the operating system and resource-management-enabled parallel runtime systems. The resource management components and runtime systems may collectively implement dynamic co-scheduling of hardware contexts when executing multiple parallel applications, using a spatial scheduling policy that grants high priority to one application per hardware context and a temporal scheduling policy for re-allocating unused hardware contexts. The runtime systems may receive resources on a varying number of hardware contexts as demands of the applications change over time, and the resource management components may co-ordinate to leave one runnable software thread for each hardware context. Periodic check-in operations may be used to determine (at times convenient to the applications) when hardware contexts should be re-allocated. Over-subscription of worker threads may reduce load imbalances between applications.

Abstract: Multi-core computers may implement a resource management layer between the operating system and resource-management-enabled parallel runtime systems. The resource management components and runtime systems may collectively implement dynamic co-scheduling of hardware contexts when executing multiple parallel applications, using a spatial scheduling policy that grants high priority to one application per hardware context and a temporal scheduling policy for re-allocating unused hardware contexts. The runtime systems may receive resources on a varying number of hardware contexts as demands of the applications change over time, and the resource management components may co-ordinate to leave one runnable software thread for each hardware context. Periodic check-in operations may be used to determine (at times convenient to the applications) when hardware contexts should be re-allocated. Over-subscription of worker threads may reduce load imbalances between applications.

Abstract: Fast modern interconnects may be exploited to control when garbage collection is performed on the nodes (e.g., virtual machines, such as JVMs) of a distributed system in which the individual processes communicate with each other and in which the heap memory is not shared. A garbage collection coordination mechanism (a coordinator implemented by a dedicated process on a single node or distributed across the nodes) may obtain or receive state information from each of the nodes and apply one of multiple supported garbage collection coordination policies to reduce the impact of garbage collection pauses, dependent on that information. For example, if the information indicates that a node is about to collect, the coordinator may trigger a collection on all of the other nodes (e.g., synchronizing collection pauses for batch-mode applications where throughput is important) or may steer requests to other nodes (e.g., for interactive applications where request latencies are important).

Abstract: Multi-core computers may implement a resource management layer between the operating system and resource-management-enabled parallel runtime systems. The resource management components and runtime systems may collectively implement dynamic co-scheduling of hardware contexts when executing multiple parallel applications, using a spatial scheduling policy that grants high priority to one application per hardware context and a temporal scheduling policy for re-allocating unused hardware contexts. The runtime systems may receive resources on a varying number of hardware contexts as demands of the applications change over time, and the resource management components may co-ordinate to leave one runnable software thread for each hardware context. Periodic check-in operations may be used to determine (at times convenient to the applications) when hardware contexts should be re-allocated. Over-subscription of worker threads may reduce load imbalances between applications.