Rubrik, Inc.

Microsoft Technology Licensing, LLC

Daniel Hefenbrock

Jeffrey Douglas Hughes

Triantaphyllos Byron Rakitzis

Abhishek Gupta

Ran Kalach

Anshul Mohan Gupta

Ping Xie

Muraliraja Muniraju

Chun Ho Cheung

Shivanshu Agrawal

Kashif Hasan

Takeshi Suzuki

Triantaphyllos Rakitzîs

Methods, systems, and devices for data management are described. A data management system (DMS) may obtain metadata associated with a snapshot stored by the DMS and including information associated with data paths in the snapshot. The DMS may scan an index registry for identifiers of the data paths to determine whether the index registry includes namespace and hierarchical index information that map the data paths to path identifiers and to descendent data paths within the snapshot, respectively. The DMS may generate term and count index information based on the metadata. The term and count index information may map terms to data paths that include the terms and numerical values associated with the terms, respectively. The DMS may store, in the index registry, the term and count index information with an identifier of a corresponding index group. The namespace and hierarchical index information may be common to multiple index groups.

Metadata search, filtering, and aggregation using indexes

Embodiments are directed to managing file systems over a network. A hierarchical index may be provided based on a file system and a plurality of objects stored in the file system A token index may be generated based on the hierarchical index. Each token may be a portion of the path of the objects Metric indices may be generated based on the hierarchical index and a plurality of metrics associated with the objects such that the metrics indices include one or more rows that corresponds to a place position for a metric value. Employing the token index and the metric indices to generate query results based on the plurality of metrics associated with the objects.

Aggregating metrics in distributed file systems

Embodiments are directed to managing file systems. Update information associated with a change of a metric associated with a target object may be provided. A journal that includes a base bin that includes base records that associate the metric with each object in the file system. Records that include the change of the metric associated with the ancestors of target object may be generated. Another record that includes the change of the metric associated with the target object may be generated. A level bin associated with the base bin of the journal may be provided based on the update information. The records may be stored in the level bin using a sort order based on the ordering of the base bin records. In response a query, the journal may be employed to reduce latency in generating query results.

Aggregating metrics in file systems using structured journals

The subject disclosure is directed towards partially recalling file ranges of deduplicated files based on tracking dirty (write modified) ranges (user writes) in a way that eliminates or minimizes reading and writing already-optimized adjacent data. The granularity of the ranges does not depend on any file-system granularity for tracking ranges. In one aspect, lazy flushing of tracking data that preserves data-integrity and crash-consistency is provided. In one aspect, also described is supporting granular partial recall on an open file while a data deduplication system is optimizing that file.

Granular partial recall of deduplicated files

Growing community of inventors

Palo Alto, CA, United States of America

Daniel Hefenbrock