Robin Systems, Inc.

Dhanashankar Venkatesan

Jagadish Kumar Mukku

Ripulkumar Hemantbhai Patel

Partha Sarathi Seetala

Ravi Kumar Alluboyina

Shravan Kumar Vallala

Gurmeet Singh

Krishna Satyasai Yeddanapudi

Manjunath Mageswaran

Kallur Vasudeva Rao Narasimha Subban

Rafit Izhak Ratzin

A new snapshot of a storage volume is created by instructing computing nodes to suppress write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot and finalizes and performs garbage collection with respect to segments allocated to the previous snapshot. Subsequent write requests to the storage volume are then performed on the segments allocated to the new snapshot. A segment maps segments to a particular snapshot and metadata stored in the segment indicates storage volume addresses of data written to the segment. The snapshots may be represented by a storage manager in a hierarchy that identifies an ordering of snapshots and branches to clone snapshots.

Creating snapshots of a storage volume in a distributed storage system

A new snapshot of a storage volume is created by suppressing write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot. Subsequent write requests to the storage volume are then performed on the segments allocated to the new snapshot. An orchestration layer implements a bundled application that is provisioned with storage volumes and containers. A snapshot of the application may be created and used to rollback or clone the application. De-duplication may be performed by creating a signature map and identifying duplicated blocks. Blocks of segments containing duplicated blocks are copied to pool segments and metadata of those segments of the same logical storage unit may be consolidated to pool metadata segments. The identification of duplicate blocks may be performed in a cloud computing platform.

Data de-duplication

A new snapshot of a storage volume is created by instructing computing nodes to suppress write requests. Storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot and finalizes and performs garbage collection with respect to segments allocated to the previous snapshot. Subsequent write requests to the storage volume are then performed on the segments allocated to the new snapshot. A segment maps segments to a particular snapshot and metadata stored in the segment indicates storage volume addresses of data written to the segment. The snapshots may be represented by a storage manager in a hierarchy that identifies an ordering of snapshots and branches to clone snapshots. A non-snapshot volume may be converted to a snapshot volume at any point after creation.

A new snapshot of a storage volume is created by suppressing write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot. Data is written in segments and Metadata is stored in the segments indicating LBAs of data stored therein and offsets within the segments at which data for LBAs are stored. For write requests, index entries are stored in an index buffer for a segment and written to the segment when the buffer is full. A redo entry is created in a redo buffer for multiple storage volumes and slices for each write request. Write requests are acknowledged when the redo buffer is written to redo segments on a storage device. On restart, index buffers are reconstructed from the redo segments if needed.

Redo log for append only storage scheme

A new snapshot of a storage volume is created by suppressing write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot. Subsequent write requests to the storage volume are then performed on the segments allocated to the new snapshot. A block map records segments where current data for an LBA of a slice of a storage volume is stored. Block maps may be written to a storage device in order to free memory. Block maps may be read back into memory when needed. Writing and reading of block maps may be performed upon fragments of block maps. On restarting of the storage node, block maps may be restored from block maps stored in the storage device.

Block map cache

A distributed storage system stores a storage volume as segments that are allocated as needed and assigned VSIDs according to a monotonically increasing counter. The storage volume may be provisioned by an orchestration layer that manages the storage volumes as well as containers executing executable components of the storage volume. The storage volume may be replicated, such as by replicating slices of the storage volume. A primary copy of the slice may be moved from one node to another within the distributed storage system by designating it as a replica, creating a new replica at the new location which is then brought current. The new replica is then designated as the primary replica and the former primary replica may be deleted. A non-replicated storage volume may be converted to a replicated storage volume and vice versa.

Changing a distributed storage volume from non-replicated to replicated

A distributed storage schemes manages implementation of QoS targets for IOPs across compute nodes executing applications, primary storage nodes storing a primary copy of a logical storage volume, and clone storage nodes. On the compute node, a maximum priority is assigned to a minimum number of IOPs in a queue within a time window from a time of receipt of a last unexecuted IOP. Other IOPs are assigned a minimum priority. On the storage node, maximum priority IOPs are assigned to high priority queues, from which IOPs are executed first, and low priority IOPs are assigned to low priority queues. Methods for determining the capacity of storage nodes and allocating storage requests are also disclosed.

Allocating storage requirements in a distributed storage system

A new snapshot of a storage volume is created by suppressing write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot. Subsequent write requests to the storage volume are then performed on the segments allocated to the new snapshot. An orchestration layer implements a bundled application that is provisioned with storage volumes and containers. A snapshot of the application may be created and used to rollback or clone the application. The amount of processing cores, memory, and containers may be increased or decreased based on usage. Amount of storage allocated to a storage volume may be augmented by a snapshot reservation ratio. The snapshot reservation ratio is adjusted over time according to usage.

Snapshot reservations in a distributed storage system

A new snapshot of a storage volume is created by suppressing write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot. Subsequent write requests to the storage volume are then performed on the segments allocated to the new snapshot. An orchestration layer implements a bundled application that is provisioned with storage volumes and containers. A snapshot of the application may be created and used to rollback or clone the application. The amount of processing cores, memory, and containers may be increased or decreased based on usage. Components of the bundled application may be assigned to nodes to satisfy affinity and anti-affinity rules. Device sets may be generated according to these rules and used to plan for the assignment of storage volumes and containers of the bundled application.

Implementing affinity and anti-affinity constraints in a bundled application

Window-based prority tagging of IOPs in a distributed storage system

A new snapshot of a storage volume is created by instructing computing nodes to suppress write requests. A snapshot of the application may be created and used to rollback or clone the application. Clones snapshots of storage volumes may be gradually populated with data from prior snapshots to reduce loading on a primary snapshot. Components of cloned applications may communicate with one another using addresses of these components in the parent application. Changes to application state may be written to a remote storage volume that may be mounted to a new instance or clone of the application to recreate its state. Jobs to create and execute a bundled application may be referenced with a simulated file system that generates reads to hosts only when the job log file is actually read. Storage nodes may implement interfaces to a SAN or cloud storage system.

Implementing containers for a stateful application in a distributed computing system

A distributed storage system stores a storage volume as segments that are allocated as needed and assigned VSIDs according to a monotonically increasing counter. The storage volume may be provisioned by an orchestration layer that manages the storage volumes as well as containers executing executable components of the storage volume. A storage volume may be stored across multiple tiers with a storage node managing on which tier a segment is stored based on a volume tier policy and usage of the segments. The tiering policy may be used to relocate the entire storage volume to a tier, such as a cloud computing platform, after which an application may be restored on the tier. Segments on a tier with built in redundancy may be used as both primary and backup copies.

Implementing storage volumes over multiple tiers

A storage scheme allocates portions of a logical volume to storage nodes in excess of the capacity of the storage nodes. Slices of the storage nodes and segments of slices are allocated in response to write requests such that actual allocation on the storage nodes is only in response to usage. Segments are identified with virtual segment identifiers that are retained when segments are moved to a different storage node. Logical volumes may therefore be moved seamlessly to different storage nodes to ensure sufficient storage capacity. Data is written to new locations in segments having space and a block map tracks the last segment to which data for a given address is written. Garbage collection is performed to free segments that contain invalid data, i.e. data for addresses that have been subsequently written to.

Storage scheme for a distributed storage system

A new snapshot of a storage volume is created by instructing computing nodes to suppress write requests. Once pending write requests from the computing nodes are completed, storage nodes create a new snapshot for the storage volume by allocating a new segment to the new snapshot and finalizes and performs garbage collection with respect to segments allocated to the previous snapshot. The snapshots may be represented by a storage manager in a hierarchy. Deleted snapshots may be flagged as such in the hierarchy and deletion may be implemented only in memory on a storage node, which is then restored from the hierarchy in the event of a crash. A snapshot is removed from the hierarchy when all segments previously are freed by garbage collection. A hybrid storage node may perform both computing and storage services. Data may be written with tags indicating encoding protocols used to encode the data.

Growing community of inventors

San Jose, CA, United States of America

Dhanashankar Venkatesan