摘要:
A computer-implemented method for managing sub-clusters within a multi-cluster computing system subsequent to a network-partition event may include: 1) identifying a plurality of computer clusters, 2) identifying a network-partition event that partitions each of the plurality of computer clusters into multiple sub-clusters, 3) identifying, within the multiple sub-clusters, a plurality of nodes that share a common attribute or property, and then 4) ensuring that only those nodes that share the common attribute survive subsequent to the network-partition event. Various other methods, systems, and computer-readable media are also disclosed.
摘要:
When a loss of connectivity between a first node in a cluster of nodes and a data store is detected, an input/output (I/O) request associated with an application that was executing on the first node is shipped to a second node in the cluster that is in communication with the data store. The application can be gracefully shut down on the first node, and the second node can execute the application and satisfy the I/O request.
摘要:
A computer-implemented method for natural batching of I/O operations on a replication log may include: 1) identifying a replication log that records the order of writes within a cluster replication system, 2) determining that the replication log is unavailable, 3) queuing incoming I/O operations for the replication log in a single batch while the replication log is unavailable, 4) determining that the replication log has become available, 5) ceasing queuing of incoming I/O operations for the replication log based on the determination that the replication log has become available, and 6) grouping a plurality of I/O operations in the single batch for processing in parallel by assigning a same generation number to the plurality of I/O operations. Various other methods, systems, and computer-readable media are also disclosed.
摘要:
When a loss of connectivity between a first node in a cluster of nodes and a data store is detected, an input/output (I/O) request associated with an application that was executing on the first node is shipped to a second node in the cluster that is in communication with the data store. The application can be gracefully shut down on the first node, and the second node can execute the application and satisfy the I/O request.
摘要:
A computer-implemented method for efficient sequential logging on caching-enabled storage devices may include 1) identifying a storage device with a cache, 2) allocating space on the storage device for a sequential log, 3) calculating a target size for the sequential log based at least in part on an input/output load directed to the sequential log, and then 4) restricting the sequential log to a portion of the allocated space corresponding to the target size. Various other methods, systems, and computer-readable media are also disclosed.
摘要:
A computer-implemented method for efficient sequential logging on caching-enabled storage devices may include 1) identifying a storage device with a cache, 2) allocating space on the storage device for a sequential log, 3) calculating a target size for the sequential log based at least in part on an input/output load directed to the sequential log, and then 4) restricting the sequential log to a portion of the allocated space corresponding to the target size. Various other methods, systems, and computer-readable media are also disclosed.