Rubrik, Inc.

Google Inc.

Hyo Jun Kim

Pin Zhou

Prasenjit Sarkar

Maohua Lu

Ajaykrishna Raghavan

Rohit Shekhar

Rohit Upadhyaya Jayasankar

Embodiments disclosed herein provide systems, methods, and computer readable media for sub-cluster recovery in a data storage environment having a plurality of storage nodes. In a particular embodiment, the method provides scanning data items in the plurality of nodes. While scanning, the method further provides indexing the data items into an index of a plurality of partition groups. Each partition group includes data items owned by a particular one of the plurality of storage nodes. The method then provides storing the index.

Sub-cluster recovery using a partition group index

Example embodiments of the present disclosure provide, in one example aspect, an example computer-implemented method for verification of a shared cache. The example method can include retrieving a precompiled shared cache entry corresponding to a shared cache key, the shared cache key being associated with an operation request. The example method can include obtaining a directly compiled resource associated with the operation request. The example method can include certifying one or more portions of the shared cache based at least in part on a comparison of the precompiled shared cache entry and the directly compiled resource.

Shared compilation cache verification system

The method disclosed is for instantiating a second cluster based on a first cluster. For at least one node of a second plurality of nodes, generating per node data based on mappings between a plurality of partition groups and a first plurality of nodes, the first plurality of nodes corresponding to the first cluster. The method further discloses identifying data items included in the plurality of partition groups based on the mappings between the plurality of partition groups and the first plurality of nodes. The method further discloses each partition group corresponding to a node of the first plurality of nodes and comprising a subset of data items stored in the node. The method further discloses loading the data items included in the plurality of partition groups onto the second plurality of nodes, the second plurality of nodes corresponding to the second cluster.

The method disclosed includes scanning data items stored in the first plurality of nodes of a first cluster. While scanning, creating a partition group index indexing the data items into a plurality of partition groups. Each partition group corresponds to a node of the first plurality of nodes and comprises a subset of data items stored in the node. Storing the index. Instantiating a second cluster, comprising generating per node data, for each node of a second plurality of nodes, based on mappings between the partition groups and the first plurality of nodes. Identifying the data items included in the partition groups according to the partition group index and loading the data items included in the partition groups onto the second plurality of nodes.

Embodiments disclosed herein provide systems, methods, and computer readable media for generating a consistent deduplicated snapshot of a distributed database using optimistic deduplication. In a particular embodiment, a method provides, for each node of a plurality of nodes in the distributed database, deduplicating data items stored on the node that are owned by the node and generating a summary that describes a file in which the data items are located. The method further provides identifying from the summaries for each of the nodes whether mistakes occurred during deduplication and, upon identifying one or more mistakes, determining one or more corrections for the one or more mistakes. Also, the method provides generating a consistent deduplicated snapshot for the distributed database comprising the deduplicated data items from each node and the one or more corrections.

Growing community of inventors

San Jose, CA, United States of America

Hyo Jun Kim