Druva Inc.

Symantec Operating Corporation

Veritas Operating Corporation

Druva Technologies Pte. Ltd.

Milind Vithal Borate

Somesh Jain

Anand Apte

Aalop Suresh Shah

Shekhar S Deshkar

Basant Rajan

Samir Desai

Murali Nagaraj

Prahlad Nishal

Yogendra Acharya

Angshuman Bezbaruah

Veeral P Shah

Ganesh Varadarajan

Craig Harmer

Peter Vajgel

Sree Hari Nagaralu

Amit Shirish Kulkarni

Shriram Wankhade

Srikiran Gottipati

Alok Kumar

Gaurav Gangalwar

Ravideep Singhvi

Pallavi Thakur

Aditya Kelkar

Jaspreet Singh

Aditya Agrawal

Anup Agarwal

Nikhil Singhal

John Colgrove

Oleg Kiselev

John A Muth

Ajay Potnis

Sudhakar Paulzagade

Nishant Shivajirao Thorat

Sudeep Jathar

Shubham Agarwal

Nishith Chitaliya

Nookala Sunder Phani Kumar

Bhimsen Bhanjois

Neeraj Thakur

Rohit Singh

George Manianghat Mathew

Basant A/l R P Rajan

Pinkesh Bardiya

Trimbak Bardale

Vinay Punera

Faisal Puthuparackat

A file system and a related method are presented. The file system includes a data storage including a plurality of data blocks; a merge index including a plurality of namespace entries, wherein the plurality of namespace entries include a plurality of blockmap entries and a plurality of local reference entries; a deduplication database including a plurality of deduplication indices and a plurality of global reference entries for a plurality of datasets; and an indexing system configured to generate the plurality of namespace entries and the plurality of global reference entries.

Systems and method for indexing and data compaction

Data compaction systems and method

A data management system manages concurrent readers and writers for large file scans. The data management system may read a plurality of data chucks of the file starting from different offsets and generate a bounded number of read requests, which causes a data chuck identifiable by a data offset to be loaded into a data buffer. The system may queue the loaded data chucks for generating write requests to release the loaded data chunks. One or more write requests are generated responsive to one or more data chunks being associated with a consecutive order of data offsets being successfully loaded to data buffers. The system may write data chucks released from the buffer-rounded reading stage to the data storage in a checkpointed writing stage. The checkpointed writing stage creates a checkpoint based on the data offset of the data chucks that have been completely transferred to the data storage.

Data management system for transferring files between databases

System for delivering an event journal during a back-up session in a distributed file system is presented. The system includes an event intake module, a load balancer, a plurality of object creation modules, a journal manager, and a journal service module. Each object creation module of the plurality of object creation modules further includes an ingestor and a drainer. A related method is also presented. The system and method provide for reliable and time-ordered delivery of events in the event journal.

System and method for event journal delivery

A data back-up system configured to back-up one or more data sets from one or more devices to a data back-up server by using a proxy pool is presented. The data back-up system includes a load-balancer configured to distribute the one or more data sets across the proxy pool. The load balancer includes a data receiver configured to receive types of files, number of files, and total size of each file in the one or more data sets; a load estimator configured to estimate a weighted average load of each data set based on the number of files, the total size of each file, a compressibility factor for each file type, and an encryption factor for each file type; and a load distributor configured to distribute the one or more data sets as a plurality of workloads across the proxy pool.

System and method for data back-up using a proxy pool

A folder scan system configured to identify modified folders in a storage module including a plurality of folders during an incremental backup scan is presented. The folder scan system is configured to identify modified folders using a learning-based technique. A related method is also presented.

Folder scan system and method for incremental backup

A system includes a data store and a data management server. The data store stores a plurality of backup snapshots that capture states of a device at different times. The data management server receives a request to restore the device that is potentially malware affected. The data management server retrieves a first backup snapshot of the device. The data management server determines that a first file stored in the first backup snapshot is malware affected. The data management server checks one or more corresponding versions of the first file captured in one or more previous backup snapshots to identify a clean version of the first file. The data management server determines that a second file stored in the first backup snapshot is clean. The data management server restores data in the device.

Recovery point objective optimized file recovery

A backup management system may include a data management server, a warm-tier data store, and a cold-tier data store. Snapshots may be captured from various client devices. A data block stored in the warm-tier data store may be referenced by multiple backup snapshots and/or referenced by one or more users. When a data block's total reference count is equal to the cold reference count or equal to or less than a threshold total reference count, the data management server may determine that the data block is ready to be migrated to the cold-tier data store. The data management server may send the data block into a candidate queue. In the queue, data blocks with similar retention periods or similar expected restoration may be grouped as a unit. The unit may be transmitted to the cold-tier data store in a single write request.

Storage tiering for backup data

A system and a method for on-demand search of a large data-set is presented. The system includes a data indexer, an index writer, and an index reader. The data indexer is configured to index the data set. The index writer is configured to create a multi-level directory including a plurality of directories having one or more hash partitions. The index writer is further configured to generate a hash table for each directory and write data from the indexed data set into a corresponding hash partition of a directory. The index reader is configured to identify and query a hash partition in each directory based on a search term and a corresponding hash table for the directory. The index reader is further configured to retrieve one or more relevant records, and present the one or more relevant records to a user.

System and method for on-demand search of a large dataset

A system for estimating one or more data storage parameters and/or statistics in a data storage system is presented. The data storage system includes a plurality of storage containers. The system includes a snapshot module, a container stats aggregator, a synchronization module, a global stats aggregator, and storage stats estimator.

Event based aggregation for distributed scale-out storage systems

A file system may include an object storage, a merged index, and a distributed database. When a file is stored in the file system, the file may be converted to an object and be stored in the object storage. The deduplication index of the file may be stored in the distributed database. The namespace metadata of the file may be stored in the merged index. The merged index generates namespace entries of the file when the file is created, deleted, and/or modified. A namespace entry may be associated with a specific file and may include a creation version and a deletion version. When a file is deleted or modified, instead of modifying the existing namespace entries, new entries associated with different versions and including different creation or deletion versions are created. The status of a file may be monitored by one or more entries associated with a file.

Deduplicated merged indexed object storage file system

Disclosed embodiments include a method (system and non-transitory computer-readable medium) for backing up updated portions of a plurality files having hierarchical relationships through object storage. In one or more embodiments, a file is segregated into chunks, and objects corresponding to the chunks are generated for storage at an object storage. For a chunk, an object for storing the chunk and additional objects for storing mapping information are generated. The mapping information may include path information identifying a path of the file in a hierarchical structure, a file version list identifying a version of the file, a chunk list describing an association between the file and the chunks, a chunk version list identifying a version of the chunk, etc. When a portion of the file is updated, objects corresponding to the updated portion of the file can be generated, and stored at the object storage.

Synthetic full backup storage over object storage

A system, method, and non-transitory computer-readable medium provide backup and archive services for computing devices. Typically, multiple backup snapshots are archived together in each archive cycle. A catalogue for the current archive is efficiently created by starting with a copy of the previous archive catalogue and updating it based on metadata associated with the backup snapshots.

Unified time-indexed catalogue for multiple archived snapshots

A data storage system allows data to be encrypted and de-duplicated at the same system. By way of example, a server of the data storage system may request a client device which intends to upload a data block to transmit a first fingerprint of the data block to the server. The first fingerprint may be derived from the plaintext of the data block. The server may apply a one-way function to the first fingerprint to generate an encryption key and transmit the encryption key to the client device. The client device uses the encryption key to encrypt the data block and generates a second fingerprint which is derived from the ciphertext of the data block. The server uses both the first fingerprint and the second fingerprint to verify the data block and the legitimacy of the client attempting to upload the data block.

Data de-duplication among untrusted entities

Growing community of inventors

Pune, India

Milind Vithal Borate