Patents by Inventor Manish Swaminathan
Manish Swaminathan has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 11392469Abstract: The disclosed embodiments provide a system for testing machine learning workflows. During operation, the system obtains a configuration for a staging test of a machine learning model, wherein the configuration includes a model name for the machine learning model, a duration of the staging test, and a use case associated with the machine learning model. Next, the system selects a staging test host for the staging test. The system then deploys the staging test on the staging test host in a staging environment, wherein the deployed staging test executes the machine learning model based on live traffic received from a production environment. After the staging test has completed, the system outputs a set of metrics representing a system impact of the machine learning model on the staging test host.Type: GrantFiled: June 20, 2019Date of Patent: July 19, 2022Assignee: Microsoft Technology Licensing, LLCInventors: Ali Sadiq Mohamed, Manish Swaminathan, Shunlin Liang, Prateek Sachdev, Vivek Desai, Adam R. Peck, Sunny Sanjiv Ketkar
-
Patent number: 11334566Abstract: Methods, systems, and computer programs are presented for reducing latency for providing a user feed containing one or more posts. One method includes operations for receiving a request to access the user feed and for performing a first query to search posts. The first query uses a first time horizon delimiting a creation time of posts and a first maximum number of posts selected for ranking. The posts from the first query are sent to the client device for presentation on a user interface. Further, a second query is performed to search posts, where the second query uses a second time horizon that is greater than the first time horizon and a second maximum number of posts for ranking that is greater than the first maximum number of posts. The posts from the first query and the second query are merged and sent to the client device for presentation.Type: GrantFiled: January 20, 2020Date of Patent: May 17, 2022Assignee: Microsoft Technology Licensing, LLCInventors: Manish Swaminathan, Manas Haribhai Somaiya, Vivek Yogesh Tripathi, Strahinja Markovic, Ali Mohamed, Muhammad Hassan Khan, Xin Hu, Caitlin Marie O'Connor, Zeesha Currimbhoy, Shunlin Liang, Prateek Sachdev, Madhulekha Arunmozhi
-
Publication number: 20210224274Abstract: Methods, systems, and computer programs are presented for reducing latency for providing a user feed containing one or more posts. One method includes operations for receiving a request to access the user feed and for performing a first query to search posts. The first query uses a first time horizon delimiting a creation time of posts and a first maximum number of posts selected for ranking. The posts from the first query are sent to the client device for presentation on a user interface. Further, a second query is performed to search posts, where the second query uses a second time horizon that is greater than the first time horizon and a second maximum number of posts for ranking that is greater than the first maximum number of posts. The posts from the first query and the second query are merged and sent to the client device for presentation.Type: ApplicationFiled: January 20, 2020Publication date: July 22, 2021Inventors: Manish Swaminathan, Manas Haribhai Somaiya, Vivek Yogesh Tripathi, Strahinja Markovic, Ali Mohamed, Muhammad Hassan Khan, Xin Hu, Caitlin Marie O'Connor, Zeesha Currimbhoy, Shunlin Liang, Prateek Sachdev, Madhulekha Arunmozhi
-
Publication number: 20200401491Abstract: The disclosed embodiments provide a system for testing machine learning workflows. During operation, the system obtains a configuration for a staging test of a machine learning model, wherein the configuration includes a model name for the machine learning model, a duration of the staging test, and a use case associated with the machine learning model. Next, the system selects a staging test host for the staging test. The system then deploys the staging test on the staging test host in a staging environment, wherein the deployed staging test executes the machine learning model based on live traffic received from a production environment. After the staging test has completed, the system outputs a set of metrics representing a system impact of the machine learning model on the staging test host.Type: ApplicationFiled: June 20, 2019Publication date: December 24, 2020Applicant: Microsoft Technology Licensing, LLCInventors: Ali Sadiq Mohamed, Manish Swaminathan, Shunlin Liang, Prateek Sachdev, Vivek Desai, Adam R. Peck, Sunny Sanjiv Ketkar
-
Patent number: 10762070Abstract: A technique reduces an amount of metadata stored in a memory of a node in a cluster. An extent store layer of a storage input/output (I/O) stack executing on the node stores key-value pairs in a plurality of data structures, e.g., cuckoo hash tables, resident in the memory. The cuckoo hash table embodies metadata that describes an extent and, as such, may be organized to associate a location on disk with a value that identifies the location on disk. The value may be embodied as a locator that includes a reference count used to support deduplication functionality of the extent store layer with respect to the extent. The reference count is divided into two portions: a delta count portion stored in memory for each slot of the hash table and an overflow count portion stored on disk in a header of each extent. One bit of the delta count portion is reserved as an overflow bit that indicates whether the in-memory reference count has overflowed.Type: GrantFiled: February 13, 2018Date of Patent: September 1, 2020Assignee: NetApp, Inc.Inventors: Manish Swaminathan, Dhaval Patel, Edward D. McClanahan, Jeffrey S. Kimmel
-
Patent number: 10210082Abstract: A rate matching technique may be configured to adjust a rate of cleaning of one or more selected segments of the storage array to accommodate a variable rate of incoming workload processed by a storage input/output (I/O) stack executing on one or more nodes of a cluster. An extent store layer of the storage I/O stack may clean a segment in accordance with segment cleaning which, illustratively, may be embodied as a segment cleaning process. The rate matching technique may be implemented as a feedback control mechanism configured to adjust the segment cleaning process based on the incoming workload. Components of the feedback control mechanism may include one or more weight schedulers and various accounting data structures, e.g., counters, configured to track the progress of segment cleaning and free space usage. The counters may also be used to balance the rates of segment cleaning and incoming I/O workload, which may change depending upon an incoming I/O rate.Type: GrantFiled: April 28, 2017Date of Patent: February 19, 2019Assignee: NetApp, Inc.Inventors: Dhaval Patel, Manish Swaminathan, Edward D. McClanahan, John Muth
-
Patent number: 10133511Abstract: An optimized segment cleaning technique is configured to efficiently clean one or more selected portions or segments of a storage array coupled to one or more nodes of a cluster. A bottom-up approach of the segment cleaning technique is configured to read all blocks of a segment to be cleaned (i.e., an “old” segment) to locate extents stored on the SSDs of the old segment and examine extent metadata to determine whether the extents are valid and, if so, relocate the valid extents to a segment being written (i.e., a “new” segment). A top-down approach of the segment cleaning technique obviates reading of the blocks of the old segment to locate the extents and, instead, examines the extent metadata to determine the valid extents of the old segment.Type: GrantFiled: September 12, 2014Date of Patent: November 20, 2018Assignee: NetApp, IncInventors: John Muth, Edward D. McClanahan, Dhaval Patel, Manish Swaminathan
-
Publication number: 20180173703Abstract: A technique reduces an amount of metadata stored in a memory of a node in a cluster. An extent store layer of a storage input/output (I/O) stack executing on the node stores key-value pairs in a plurality of data structures, e.g., cuckoo hash tables, resident in the memory. The cuckoo hash table embodies metadata that describes an extent and, as such, may be organized to associate a location on disk with a value that identifies the location on disk. The value may be embodied as a locator that includes a reference count used to support deduplication functionality of the extent store layer with respect to the extent. The reference count is divided into two portions: a delta count portion stored in memory for each slot of the hash table and an overflow count portion stored on disk in a header of each extent. One bit of the delta count portion is reserved as an overflow bit that indicates whether the in-memory reference count has overflowed.Type: ApplicationFiled: February 13, 2018Publication date: June 21, 2018Inventors: Manish Swaminathan, Dhaval Patel, Edward D. McClanahan, Jeffrey S. Kimmel
-
Patent number: 9934264Abstract: A technique reduces an amount of metadata stored in a memory of a node in a cluster. An extent store layer of a storage input/output (I/O) stack executing on the node stores key-value pairs in a plurality of data structures, e.g., cuckoo hash tables, resident in the memory. The cuckoo hash table embodies metadata that describes an extent and, as such, may be organized to associate a location on disk with a value that identifies the location on disk. The value may be embodied as a locator that includes a reference count used to support deduplication functionality of the extent store layer with respect to the extent. The reference count is divided into two portions: a delta count portion stored in memory for each slot of the hash table and an overflow count portion stored on disk in a header of each extent. One bit of the delta count portion is reserved as an overflow bit that indicates whether the in-memory reference count has overflowed.Type: GrantFiled: June 2, 2015Date of Patent: April 3, 2018Assignee: NetApp, Inc.Inventors: Manish Swaminathan, Dhaval Patel, Edward D. McClanahan, Jeffrey S. Kimmel
-
Publication number: 20170235673Abstract: A rate matching technique may be configured to adjust a rate of cleaning of one or more selected segments of the storage array to accommodate a variable rate of incoming workload processed by a storage input/output (I/O) stack executing on one or more nodes of a cluster. An extent store layer of the storage I/O stack may clean a segment in accordance with segment cleaning which, illustratively, may be embodied as a segment cleaning process. The rate matching technique may be implemented as a feedback control mechanism configured to adjust the segment cleaning process based on the incoming workload. Components of the feedback control mechanism may include one or more weight schedulers and various accounting data structures, e.g., counters, configured to track the progress of segment cleaning and free space usage. The counters may also be used to balance the rates of segment cleaning and incoming I/O workload, which may change depending upon an incoming I/O rate.Type: ApplicationFiled: April 28, 2017Publication date: August 17, 2017Inventors: Dhaval Patel, Manish Swaminathan, Edward D. McClanahan, John Muth
-
Patent number: 9671960Abstract: A rate matching technique may be configured to adjust a rate of cleaning of one or more selected segments of the storage array to accommodate a variable rate of incoming workload processed by a storage input/output (I/O) stack executing on one or more nodes of a cluster. An extent store layer of the storage I/O stack may clean a segment in accordance with segment cleaning which, illustratively, may be embodied as a segment cleaning process. The rate matching technique may be implemented as a feedback control mechanism configured to adjust the segment cleaning process based on the incoming workload. Components of the feedback control mechanism may include one or more weight schedulers and various accounting data structures, e.g., counters, configured to track the progress of segment cleaning and free space usage. The counters may also be used to balance the rates of segment cleaning and incoming I/O workload, which may change depending upon an incoming I/O rate.Type: GrantFiled: September 12, 2014Date of Patent: June 6, 2017Assignee: NetApp, Inc.Inventors: Dhaval Patel, Manish Swaminathan, Edward D. McClanahan, John Muth
-
Publication number: 20160357743Abstract: A technique reduces an amount of metadata stored in a memory of a node in a cluster. An extent store layer of a storage input/output (I/O) stack executing on the node stores key-value pairs in a plurality of data structures, e.g., cuckoo hash tables, resident in the memory. The cuckoo hash table embodies metadata that describes an extent and, as such, may be organized to associate a location on disk with a value that identifies the location on disk. The value may be embodied as a locator that includes a reference count used to support deduplication functionality of the extent store layer with respect to the extent. The reference count is divided into two portions: a delta count portion stored in memory for each slot of the hash table and an overflow count portion stored on disk in a header of each extent. One bit of the delta count portion is reserved as an overflow bit that indicates whether the in-memory reference count has overflowed.Type: ApplicationFiled: June 2, 2015Publication date: December 8, 2016Inventors: Manish Swaminathan, Dhaval Patel, Edward D. McClanahan, Jeffrey S. Kimmel
-
Publication number: 20160077746Abstract: An optimized segment cleaning technique is configured to efficiently clean one or more selected portions or segments of a storage array coupled to one or more nodes of a cluster. A bottom-up approach of the segment cleaning technique is configured to read all blocks of a segment to be cleaned (i.e., an “old” segment) to locate extents stored on the SSDs of the old segment and examine extent metadata to determine whether the extents are valid and, if so, relocate the valid extents to a segment being written (i.e., a “new” segment). A top-down approach of the segment cleaning technique obviates reading of the blocks of the old segment to locate the extents and, instead, examines the extent metadata to determine the valid extents of the old segment.Type: ApplicationFiled: September 12, 2014Publication date: March 17, 2016Inventors: John Muth, Edward D. McClanahan, Dhaval Patel, Manish Swaminathan
-
Publication number: 20160077745Abstract: A rate matching technique may be configured to adjust a rate of cleaning of one or more selected segments of the storage array to accommodate a variable rate of incoming workload processed by a storage input/output (I/O) stack executing on one or more nodes of a cluster. An extent store layer of the storage I/O stack may clean a segment in accordance with segment cleaning which, illustratively, may be embodied as a segment cleaning process. The rate matching technique may be implemented as a feedback control mechanism configured to adjust the segment cleaning process based on the incoming workload. Components of the feedback control mechanism may include one or more weight schedulers and various accounting data structures, e.g., counters, configured to track the progress of segment cleaning and free space usage. The counters may also be used to balance the rates of segment cleaning and incoming I/O workload, which may change depending upon an incoming I/O rate.Type: ApplicationFiled: September 12, 2014Publication date: March 17, 2016Inventors: Dhaval Patel, Manish Swaminathan, Edward D. McClanahan, John Muth