Microsoft - Redmond, WA, USA since Mar 2012
Principal Development Lead
Education:
University of Florida 1996 - 1998
Indian Institute of Science 1991 - 1993
Skills:
Cluster Development C++ Computer Science Databases Distributed Systems Scalability Software Design Object Oriented Design T-SQL Software Development C Cloud Computing Win32 Microsoft SQL Server Win32 API Algorithms
Interests:
database management systems, clusters, high availability, file systems, operating systems
Arkadi Brjazovski - Redmond WA, US Rohan Kumar - Redmond WA, US Sameet H. Agarwal - Redmond WA, US Stefan R. Steiner - Sammamish WA, US Mahesh K. Sreenivas - Sammamish WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 7/00
US Classification:
707792, 707802, 707812, 7079991
Abstract:
An operating system is provided. The system includes an agent component to monitor computer activities between one or more single-item access components and one or more set-based access components. A protocol component is employed by the agent component to mitigate data access conflicts between the single-item access components and the set-based access components.
Mahesh K. Sreenivas - Sammamish WA, US Benjamin Leon Grol-Prokopczyk - Redmond WA, US Jack S. Richins - Bothell WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 17/30
US Classification:
707756, 707736, 707822
Abstract:
When rich data is tunneled, its associated link identified by a primary data reference is changed by swapping primary references of original data with changed data. Thus, without duplicating the rich data, it is now associated with the changed data, preserving it in constant time. The primary references can include, but are not limited to, names and/or other item identifications. This technique allows preservation of rich data to occur without burdening processing and/or time constraints based on the size and/or number of rich data elements. The technique also allows disparate applications and/or systems to maintain data integrity even when rich data is not expected for a given application and/or system. Since the rich data follows a primary reference, awareness of the rich data is not required for it to be properly handled, ensuring its integrity.
Mahesh K. Sreenivas - Sammamish WA, US Stefan R. Steiner - Sammamish WA, US Arkadi Brjazovski - Redmond WA, US Sameet H. Agarwal - Redmond WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 12/00
US Classification:
707822
Abstract:
A file namespace hierarchy bypass and method to access files are provided. The bypass can be employed, for example, as a computer-readable medium with instructions for generating a unique identifier associated with a file and instructions for accessing the file based on the unique identifier and a specified search space. A computer-readable medium having instructions for obtaining the unique identifier and specified search space associated with the file and instructions for generating a path designation by combining the unique identifier and the specified search space used to access the file may also be employed. The method for bypassing the namespace hierarchy of the file, for example, may include passing the unique identifier and specified search space to a program and accessing the file based on a path designation generated by the program by combining the unique identifier and specified search space.
Arkadi Brjazovski - Redmond WA, US Rohan Kumar - Redmond WA, US Sameet H. Agarwal - Redmond WA, US Stefan R. Steiner - Sammamish WA, US Mahesh K. Sreenivas - Sammamish WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 7/00 G06F 12/00
US Classification:
707703, 707694, 707999201, 707792, 711100
Abstract:
An operating system is provided. The system includes an agent component to monitor computer activities between one or more single-item access components and one or more set-based access components. A protocol component is employed by the agent component to mitigate data access between the single-item access components and the set-based access components.
User Mode File System Serialization And Reliability
Arkadi Brjazovski - Redmond WA, US Rohan Kumar - Redmond WA, US Cristian Diaconu - Kirkland WA, US Oleksandr Gololobov - Redmond WA, US Mahesh K. Sreenivas - Sammamish WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 3/00 G06F 9/44 G06F 9/46 G06F 13/00
US Classification:
719319
Abstract:
An operating system is provided. The system includes an agent component to monitor computer activities between one or more single-item access components (SIAC) and one or more set-based access components (SBAC). An interface component is employed by the agent component to process kernel mode requests from a user mode application when communicating with a storage platform. Re-try components can be provided to facilitate cooperation between the SIAC and the SBAC.
Mechanism To Exchange Primary Data Stream Of A File
Mahesh Sreenivas - Sammamish WA, US Arkadi Brjazovski - Redmond WA, US Stefan Steiner - Sammamish WA, US Rohan Kumar - Redmond WA, US Sameet Agarwal - Redmond WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 17/30
US Classification:
707202000
Abstract:
A file stream exchange system and method are provided. The system can be employed, for example, with a storage system that supports transactions as a primary mechanism and employs a rich type system. The system provides a mechanism for the storage system to provide reliable updates to file streams while preserving the common criteria for security. The system employs a unique way of remapping the file-streams associated with the file-backed items that eliminates the need for copying over basic attributes, security information and timestamps etc. Significantly, the system does not rename the files involved in the operation nor does the system rely on attribute tunneling to copy over any subset of attributes for the completion of the task. The system includes a replace stream component which is a kernel mode mechanism that allows transactional exchange of stream data of two file-backed items without violating the common criteria for security.
Mahesh K. Sreenivas - Sammamish WA, US Robert H. Gerber - Bellevue WA, US Vishal Kathuria - Woodinville WA, US John F. Ludeman - Sammamish WA, US Ashwin Shrinivas - Sammamish WA, US Michael A. Uhlar - Seattle WA, US
Assignee:
Microsoft Corporation - Redmond WA
International Classification:
G06F 17/30
US Classification:
707641, 707E17005
Abstract:
A technique is described for storing data from a database across a plurality of data storage devices, wherein each data storage device is capable of being accessed only by a corresponding computer system in a group of interconnected computer systems. In accordance with the technique, an identifier of the database is received. An identifier of a storage tier instance is also received, wherein the storage tier instance comprises a logical representation of one or more storage locations within each of the data storage devices. Responsive to the receipt of the identifier of the database and the identifier of the storage tier instance, data from the database is stored in two or more of the storage locations logically represented by the storage tier instance, wherein each of the two or more storage locations in which data is stored is within a corresponding one of the data storage devices.
Data Placement Transparency For High Availability And Load Balancing
Vishal Kathuria - Woodinville WA, US Robert H. Gerber - Bellevue WA, US Mahesh K. Sreenivas - Sammamish WA, US Yixue Zhu - Sammamish WA, US John Ludeman - Redmond WA, US Ashwin Shrinivas - Sammamish WA, US Ming Chuan Wu - Bellevue WA, US
Assignee:
MICROSOFT CORPORATION - Redmond WA
International Classification:
G06F 17/30
US Classification:
707674, 707E17002
Abstract:
A method of updating a clone data map associated with a plurality of nodes of a computer system is disclosed. The clone data map includes node identification data and clone location data. A node failure event of a failed node of the computer system that supports a primary clone is detected. The clone data map is updated such that a secondary clone stored at a node other than the failed node is marked as a new primary clone. In addition, clone data maps may be used to perform node load balancing by placing a substantially similar number of primary clones on each node of a node cluster or may be used to increase or decrease a number of nodes of the node cluster. Further, data fragments that have a heavy usage or a large fragment size may be reduced in size by performing one or more data fragment split operations.