Abstract: Load balancing VM selection and movement is disclosed, including: selecting a combined predicted metric to be improved with respect to a plurality of storage devices; and determining a set of VMs to be moved from a first storage device included in the plurality of storage devices to a second storage device included in the plurality of storage devices such that the selected combined metric to improve is predicted to improve.

Abstract: File system object storage is disclosed, including: receiving, via a communication interface, a request to perform a file system operation; determining a file system object associated with the request, wherein the file system object comprises two or more files; and performing the file system operation in a manner determined at least in part by data associated with the file system object.

Abstract: Synchronous replication end to end latency reporting is disclosed. In various embodiments, performance data associated with processing and replicating synchronously file system operation requests from a primary storage system to a secondary storage system is received from a data storage system via a communication interface. Performance data associated with an application-level workload is aggregated. A representation of the aggregated performance data for the application-level workload is provided via a user interface.

Abstract: Offloading techniques to assist a hypervisor with VM data migration between storage devices are disclosed. In an embodiment, a method to offload live data migration from hypervisors includes migrating snapshots from a source device to a destination device, where the snapshots are associated with a virtual machine (VM) being migrated from a source device to a destination device. The method includes requesting a hypervisor to take a hypervisor snapshot. The method includes taking a storage array-based snapshot after requesting the hypervisor to take the hypervisor snapshot, and requesting the hypervisor to migrate live data from the source device to the destination device including by using the hypervisor snapshot to offload hypervisor live data migration. Hypervisor offload can be performed for various hypervisors and protocols including vSphere Storage APIs Array Integration (VAAI) and Remote offloaded data transfer (ODX).

Abstract: VM data migration between storage devices is disclosed. In an embodiment, a method of VM data migration includes determining snapshot data and live data corresponding to the VM based on a request to migrate a VM from a source device to a destination device. The method migrates the snapshot data from the source device to the destination device including by setting up a migration path and creating a migration configuration in the source device. The method requests a hypervisor to migrate the live data from the source device to the destination device. The method removes the migration path, the migration configuration, and snapshot data from the source device.

Abstract: Modeling space consumption of a migrated VM is disclosed, including: obtaining aggregated effective historical space usage data associated with a plurality of VMs at a storage device; using the aggregated effective historical space usage data to generate one or more predicted metrics of space usage at the storage device; and using the one or more predicted metrics of space usage to potentially migrate a VM with respect to the storage device.

Abstract: Efficient snapshot restore is disclosed. An indication to restore a snapshot from a replication destination to a storage system is received. At least a subset of data associated with the snapshot that is already present at the storage system is determined. A restore operation of the snapshot is performed by obtaining the at least subset of the data associated with the snapshot that is already present at the storage system locally from the storage system and by obtaining the remaining data associated with the snapshot from the replication destination.

Abstract: Load balancing VM selection and movement is disclosed, including: selecting a combined predicted metric to be improved with respect to a plurality of storage devices; and determining a set of VMs to be moved from a first storage device included in the plurality of storage devices to a second storage device included in the plurality of storage devices such that the selected combined metric to improve is predicted to improve.

Abstract: Performing data synchronization is disclosed, including: receiving an indication to synchronize a container to a snapshot, wherein the container has a first data state and an identity, wherein the snapshot corresponds to a second data state; causing the container to have the second data state corresponding to the snapshot; and maintaining the identity of the container.

Abstract: Techniques to satisfy quality of service (QoS) requirements on a per virtual machine basis natively in a storage system are disclosed. In various embodiments, for each of a plurality of virtual machines a corresponding input/output (I/O) request queue is stored on the storage system. Requests are scheduled to be pulled from the respective request queues and added to a corresponding storage operation pipeline at the storage system in an order determined based at least in part on a per virtual machine quality of service parameter.

Abstract: In an embodiment, a system for handling a device in latency state in a redundant storage system includes a processor configured to process a write request to a plurality of devices associated with a redundant storage system. The processor is further configured to receive a set of indications of write successes from a set of devices included in the plurality of devices other than a first device, and send an indication of a completed write to a requestor associated with the write request based at least in part on an indication that the first device meets a set of one or more criteria associated with a latency state. The system further includes a memory coupled to the processor and configured to store the write request.

Abstract: An embodiment of a system for automatic data protection for virtual machines includes a processor configured to use storage attributes associated with a virtual machine to determine, for the virtual machine, a data protection priority. The processor is further configured to determine a recommendation of a data protection operation to be taken with respect to the virtual machine based at least in part on the determined data protection priority. The system further includes a memory coupled to the processor and configured to store the determined data protection priority.

Abstract: Storing data is disclosed, including: creating a clone of a snapshot comprising a first set of metadata associated with a source data, at least in part by generating for the clone a second set of metadata that includes an active index that is associated with the snapshot and is configured to store metadata associated with data values that are written to the clone; receiving a write operation to write a requested data value to the clone; and in response to the write operation, updating the active index based at least in part on the requested data value.

Abstract: Techniques to automatically allocate resources among storage system resource consumers are disclosed. In various embodiments, for each of a plurality of heterogeneous categories of resource consumer of the storage system a corresponding allocated portion of storage system resources to be available for use to perform operations associated with that category of resource consumer is determined. A storage system is configured automatically to provide access to resources of the storage system based at least in part on the respective portions of storage system resources allocated to each category of resource consumer, including by ensuring availability to workloads associated with each category at any given time at least a minimum amount of storage system resources corresponding to the portion of storage system resources allocated to that category.

Abstract: Recovery of an in-memory state in a log-structured filesystem using fuzzy checkpoints is disclosed, including: determining a portion of a data structure to checkpoint to a storage unit, wherein the structure is associated with a set of references to locations in persistent storage at which metadata is stored, wherein the portion of the data structure is dynamically determined based at least in part on a size of the data structure and a predetermined number of storage units to be associated with a checkpoint window, wherein the number of storage units to be associated with the checkpoint window is fewer than a total number of storage units associated with the persistent storage; and checkpointing the portion of the data structure to the storage unit.

Abstract: Efficient deduplication for storage systems is disclosed, including: using the current data structure to track a set of deduplication data entries associated with a collection group; determining a flush criterion has been met; and storing the set of deduplication data entries of the current data structure as one or more persistent objects associated with the collection group in a persistent storage, wherein the one or more persistent objects are usable to update a set of persistently stored metadata associated with the collection group that tracks deduplication data entries generated before a previous merge operation.

Abstract: Aggregating data associated with cluster virtual machines is disclosed, including obtaining from a plurality of storage systems a plurality of values associated with an attribute associated with a cluster virtual machine (VM); and aggregating the plurality of values associated with the attribute based at least in part on an aggregation semantic associated with the attribute.

Abstract: A system for policy-driven RAID rebuild includes an interface to a group of devices each having stripe units. At least one of the devices is a spare device available to be used in the event of a failure of a device in the group of devices. The system further includes a processor coupled to the interface and configured to determine, based at least in part on an ordering criteria, an order in which to reconstruct stripe units to rebuild a failed device in the group of devices. The processor is further configured to rebuild the failed device including by reconstructing stripe units in the determined order using the spare device to overwrite stripe units as needed.

Abstract: Performing replication of snapshots between storage systems is disclosed. A request to store at a destination system a snapshot data to represent at the destination system a state of a set of data at a first point-in-time is received. A first source system snapshot data that represents at a source system the state of the set of data at the first point-in-time depends on a second source system snapshot data that represents at the source system a state of the set of data at a second point-in-time. The snapshot data to represent at the destination system the state of the set of data at the first point-in-time is determined based on data comprising the first source system snapshot data and a destination system snapshot data that represents at the destination system a state of the set of data at a third point-in-time.

Abstract: Efficient recovery of deduplication data for high capacity systems is disclosed, including: reading from the data storage device a data structure that tracks a plurality of segments to which a plurality of persistent objects have been recently written, wherein segments are written to in a monotonically increasing numerical order; selecting a checkpoint segment from among the plurality of segments based at least in part on a plurality of segment numbers corresponding to respective ones of the plurality of segments; using the checkpoint segment and a segment associated with a latest available segment number to determine a set of candidate segments; reading at least a portion of the set of candidate segments to identify a data storage block for which a corresponding deduplication data entry is not already stored in persistently stored deduplication data entries; and storing a new deduplication data entry to insert a fingerprint associated with the data storage block in a current data structure stored in a memory