Abstract: A method for data reduction may comprise computing (i) a first sketch of a first segment and (ii) a second sketch of a second segment. The first sketch and the second sketch may each comprise a set of features that are representative of or unique to the corresponding first and second segments. The method also comprise processing the first sketch and the second sketch to generate a similarity metric indicative of whether the second segment is similar to the first segment. The method may further comprise (1) performing a differencing operation on the second segment relative to the first segment when the similarity metric is greater than or equal to a similarity threshold, or (2) storing the first segment and the second segment in a database without performing the differencing operation when the similarity metric is less than the similarity threshold.

Abstract: A method and apparatus for maintaining file system consistency in a distributed system using version vectors is presented. A method generally includes comparing incarnation and transaction identifiers of a current version vector associated with a file with incarnation and transaction identifiers of a last completed version vector associated with the file. Upon determining that a current version vector reflects operations on the file that are either earlier than or the same as the identifiers in the last completed version vector, the node performing one or more file system operations on the file.

Abstract: A method for data reduction may comprise computing (i) a first sketch of a first segment and (ii) a second sketch of a second segment. The first sketch and the second sketch may each comprise a set of features that are representative of or unique to the corresponding first and second segments. The method also comprise processing the first sketch and the second sketch to generate a similarity metric indicative of whether the second segment is similar to the first segment. The method may further comprise (1) performing a differencing operation on the second segment relative to the first segment when the similarity metric is greater than or equal to a similarity threshold, or (2) storing the first segment and the second segment in a database without performing the differencing operation when the similarity metric is less than the similarity threshold.

Abstract: A method for sketch computation is provided. The method may comprise receiving an input data stream from one or more client applications. The method may also comprise generating at least one segment from the input data stream. At least one segment may comprise a plurality of chunks. The method may further comprise computing a sketch of the at least one segment. The sketch may comprise a set of features that are representative of or unique to the at least one segment, such that the set of features corresponds to the at least one segment. The sketch may be useable for inline deduplication of at least one other input data stream received from the one or more client applications without (i) generation of a full index of the plurality of chunks or (ii) comparison of the at least one other input data stream to the full index.

Abstract: A method for data reduction may comprise computing (i) a first sketch of a first segment and (ii) a second sketch of a second segment. The first sketch and the second sketch may each comprise a set of features that are representative of or unique to the corresponding first and second segments. The method also comprise processing the first sketch and the second sketch to generate a similarity metric indicative of whether the second segment is similar to the first segment. The method may further comprise (1) performing a differencing operation on the second segment relative to the first segment when the similarity metric is greater than or equal to a similarity threshold, or (2) storing the first segment and the second segment in a database without performing the differencing operation when the similarity metric is less than the similarity threshold.

Abstract: A method for sketch computation is provided. The method may comprise receiving an input data stream from one or more client applications. The method may also comprise generating at least one segment from the input data stream. At least one segment may comprise a plurality of chunks. The method may further comprise computing a sketch of the at least one segment. The sketch may comprise a set of features that are representative of or unique to the at least one segment, such that the set of features corresponds to the at least one segment. The sketch may be useable for inline deduplication of at least one other input data stream received from the one or more client applications without (i) generation of a full index of the plurality of chunks or (ii) comparison of the at least one other input data stream to the full index.

Abstract: A method for sketch computation is provided. The method may comprise receiving an input data stream from one or more client applications. The method may also comprise generating at least one segment from the input data stream. At least one segment may comprise a plurality of chunks. The method may further comprise computing a sketch of the at least one segment. The sketch may comprise a set of features that are representative of or unique to the at least one segment, such that the set of features corresponds to the at least one segment. The sketch may be useable for inline deduplication of at least one other input data stream received from the one or more client applications without (i) generation of a full index of the plurality of chunks or (ii) comparison of the at least one other input data stream to the full index.

Abstract: A method for data processing may comprise: (a) receiving one or more input data streams from one or more client applications; (b) generating at least a first segment and a second segment from the one or more input data streams, wherein the first segment may comprise a first set of chunks and the second segment may comprise a second set of chunks; (c) computing (i) a first set of fingerprints of the first plurality of chunks and (ii) a second set of fingerprints of the second plurality of chunks; (d) processing the first set of fingerprints and the second set of fingerprints to determine that the first set of chunks and the second set of chunks meet a similarity threshold; and (e) processing the first set of chunks and the second set of chunks to determine one or more differences between the first segment and the second segment.

Abstract: A method for data reduction may comprise computing (i) a first sketch of a first segment and (ii) a second sketch of a second segment. The first sketch and the second sketch may each comprise a set of features that are representative of or unique to the corresponding first and second segments. The method also comprise processing the first sketch and the second sketch to generate a similarity metric indicative of whether the second segment is similar to the first segment. The method may further comprise (1) performing a differencing operation on the second segment relative to the first segment when the similarity metric is greater than or equal to a similarity threshold, or (2) storing the first segment and the second segment in a database without performing the differencing operation when the similarity metric is less than the similarity threshold.

Abstract: A method and apparatus for maintaining file system consistency in a distributed system using version vectors is presented. A method generally includes comparing incarnation and transaction identifiers of a current version vector associated with a file with incarnation and transaction identifiers of a last completed version vector associated with the file. Upon determining that a current version vector reflects operations on the file that are either earlier than or the same as the identifiers in the last completed version vector, the node performing one or more file system operations on the file.

Abstract: A method and apparatus for maintaining file system consistency in a distributed system using version vectors is presented. A method generally includes comparing incarnation and transaction identifiers of a current version vector associated with a file with incarnation and transaction identifiers of a last completed version vector associated with the file. Upon determining that a current version vector reflects operations on the file that are either earlier than or the same as the identifiers in the last completed version vector, the node performing one or more file system operations on the file.

Abstract: Embodiments presented herein disclose adaptive techniques for scheduling self-maintenance processes. A load predictor estimates, based on a current state of a distributed storage system, an amount of resources of the system required to perform each of a plurality of self-maintenance processes. A maintenance process scheduler estimates, based on one or more inputs, an amount of resources of the distributed system available to perform one or more of the self-maintenance processes during at least a first time period. The maintenance process scheduler determines a schedule for the one or more of the self-maintenance processes to perform during the first time period, based on the estimated amount of resources required and available.

Abstract: Embodiments presented herein disclose adaptive techniques for scheduling self-maintenance processes. A load predictor estimates, based on a current state of a distributed storage system, an amount of resources of the system required to perform each of a plurality of self-maintenance processes. A maintenance process scheduler estimates, based on one or more inputs, an amount of resources of the distributed system available to perform one or more of the self-maintenance processes during at least a first time period. The maintenance process scheduler determines a schedule for the one or more of the self-maintenance processes to perform during the first time period, based on the estimated amount of resources required and available.

Abstract: Embodiments presented herein disclose adaptive techniques for scheduling self-maintenance processes. A load predictor estimates, based on a current state of a distributed storage system, an amount of resources of the system required to perform each of a plurality of self-maintenance processes. A maintenance process scheduler estimates, based on one or more inputs, an amount of resources of the distributed system available to perform one or more of the self-maintenance processes during at least a first time period. The maintenance process scheduler determines a schedule for the one or more of the self-maintenance processes to perform during the first time period, based on the estimated amount of resources required and available.