Cohesity, Inc.

Markose Thomas

Prashant Pogde

Mohit Aron

Chinmaya Manjunath

Sunil Moolchandani

Zheng Cai

Apurv Gupta

Venkata Ranga Radhanikanth Guturi

Praveen Kumar Yarlagadda

Rupesh Bajaj

Vipin Gupta

Vinay Reddy

Akshat Agarwal

Prajakta Balkrishna Ayachit

Igor Korsunsky

Anand Arun

Sureshbabu Murugesan

Janga Varun

A request to obtain an identified content item is received. It is determined that the identified content item is stored in a plurality of locations. It is determined to obtain the identified content item from a first storage location storing a first serialized representation of a file system snapshot that includes the identified content item and a second storage location storing a second serialized representation of the file system snapshot that includes the identified content item based on available resources associated with the plurality of locations. The identified content item is extracted from the first serialized representation at the first storage location and from the second serialized representation at the second storage location including building a stubbed tree data structure using the first serialized representation and the second serialized representation. The extracted identified content item is provided using the stubbed tree data structure.

Restoration of specified content from an archive

A method and system for performing incremental backup of a network attached storage (NAS) device are described. A storage capture instance associated with a first time instance is received from a network attached storage device. At least a portion of metadata of tracked network packets associated with the network attached storage device is also received. At least one changed content item of the network attached storage device that has changed since the first time instance is identified by analyzing the at least the portion of the metadata of the tracked network packets received. An incremental backup of the network attached storage device is performed at a second time instance based at least in part on the at least one changed content item identified.

Packet-based differential backup of network-attached storage device content

A file directory tree structure of a selected storage snapshot is dynamically divided into different portions. A plurality of the different file directory tree structure portions are analyzed in parallel to identify any changes of the selected storage snapshot from a previous storage snapshot. To analyze each of the plurality of the different file directory tree structure portions, a processor is further configured to traverse and compare a corresponding file directory tree structure portion of the selected storage snapshot with a corresponding portion of a file directory tree structure of the previous storage snapshot while at least another one of the plurality of the different file directory tree structure portions of the selected storage snapshot is being analyzed in parallel.

Determining differences between two versions of a file directory tree structure

A method and apparatus for dynamically adjusting an ingestion rate for backup operations on a source system. The method generally includes monitoring a resource utilization related to one or more performance metrics of the source system in performing at least a primary workload. Based on the monitored resource utilization, the backup system determines a data ingestion rate for backup operations on the source system. The backup system ingests data from the source system to a backup repository at the determined data ingestion rate.

Dynamic throughput ingestion of backup sources

A technique for restoring content is described. The technique includes receiving a specification of content to be restored to a target storage from a data source. The technique also includes determining task units to be performed to restore the specified content. Each task unit is associated with a corresponding assigned weight value. Based on the corresponding assigned weight values, the task units are distributed to different task groups. The task groups are processed in parallel with each other to concurrently perform at least a portion of the distributed task units.

Efficient restoration of content

A method and system for obtaining identified content items that have been backed up is disclosed. The method includes receiving a request to obtain an identified content item. A backup location among a plurality of backup locations where the identified content item is stored is determined. It is determined whether the backup location corresponds to a serialized representation of a backup snapshot including the identified content item. In response to a determination that the backup location corresponds to the serialized representation, the identified content item is extracted from the serialized representation including building a stubbed snapshot tree using the serialized representation. As a response to the request, the identified content item is provided and extracted using the stubbed snapshot tree.

Techniques are disclosed for managing a high performance, fault-tolerant, strongly consistent, distributed key-value store system. The key-value store may store information, such as metadata for a distributed file system. Fault-tolerance means that the distributed key-value store continues to provide access to values in the key-value store in spite of a certain number of node failures. To provide this capability, the key-value store may store copies of (key, value) pair on N+1 nodes in order to provide fault tolerance for the failure of up to N nodes. In addition, metadata describing which nodes store a given value is stored on 2N+1 nodes and the distributed key-value store is sized such that there are 3N+1 nodes in a cluster. Doing so allows the key, value store to tolerate a failure of N nodes, while still maintaining a consistent and available key-value store.

Performing file system operations in a distributed key-value store

A method and system for migrating data are described. A storage capture instance in a first format from a first type of device is received. Processing utilized to back up the storage capture instance to a second format of a snapshot tree is distributed to multiple storage nodes. A request to migrate at least part of the storage capture instance to a second type of device in a third format is received. In response to the request, processing utilized to migrate the at least part of the storage capture instance from the second format to the second type of device in the third format is distributed to the storage nodes. In one aspect, file(s) in a first network attached storage (NAS) device format are migrated to a second NAS device format. In another aspect, a virtual machine (VM) and/or file(s) therein having a first VM format are migrated to a second VM format.

Multi-format migration for network attached storage devices and virtual machines

Distributed key-value store

A method and apparatus for scheduling backup tasks on a source system specified by a backup job to meet a recovery point objective (RPO) specified for the backup job is presented. A method generally includes estimating resource utilization at the source system expected during a plurality of time periods based on historical resource utilization by jobs executed on the source system. The backup system estimates a time required to complete the backup tasks during each time period on the source system based on the estimated resource utilization for each time period and historical backup performance data. Upon determining that the estimated time required to complete the backup tasks for one or more of the time periods meets the RPO, the backup system schedules the backup tasks to be executed during a time period of the one or more time periods with the lowest estimated resource utilization.

Growing community of inventors

San Jose, CA, United States of America

Markose Thomas