Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2001-12-19
2004-08-10
Mizrahi, Diane D. (Department: 2175)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000
Reexamination Certificate
active
06775673
ABSTRACT:
FIELD OF THE INVENTION
The present invention generally relates to distributed file systems, and more particularly to object management in distributed file systems.
BACKGROUND
In a distributed file system the data are distributed between network-connected nodes. Distributed file systems are scaleable to satisfy increasing storage and access requirements. As more storage resources are required to satisfy the requirements of various applications, new nodes with additional storage capacity can be added to the distributed file system.
The structuring of data between different nodes may impact the performance of certain applications, depending on data access patterns. For example, if one node provides access to two data sets, heavy access to the two data sets may impact the required performance level. To alleviate this type of performance bottleneck, contention for the storage resources is reduced by redistributing selected data sets amongst the different nodes.
Data are traditionally migrated at the block level, with no movement of the associated control data. Such control data includes, for example a file system directory entry, which is sometimes referred to as “meta-data”. If the control data associated with the competing data sets are not separated, the performance level may continue to be negatively impacted. Because referencing data at the block level does not indicate whether the block contains data or meta-data, it is not feasible to migrate meta-data and data blocks together.
Another approach to redistributing data is to move data at the granularity of a file system. This results in both control data and block-level data being moved. However, a file system typically includes many files that are hierarchically organized within many different directories. Thus, moving an entire file system may not separate the data sets that need to be separated. It can be seen with present systems that the granularity of moving data at the block level may be too fine, and the granularity of moving data at the file system level may be too coarse.
Migrating objects in a distributed file system also presents issues for the clients that reference the objects. One issue is the validity of cached file handles after objects have been migrated. Generally, the file handle identifies the host node and other information for referencing a file. The file handle is provided by a namespace server to a client in response to an access request, and the client caches the file handle for future references. If the data referenced by the file handle is moved, the cached file handle will be invalid. Thus, cached file handle must be revalidated for it to be useful to the client.
A system and method that address the aforementioned problems, as well as other related problems, are therefore desirable.
SUMMARY OF THE INVENTION
In various embodiments, the invention provides migration of a logical volumes in a distributed file system having multiple partitions servers. Each partition server owns one or more volumes, and each volume including a subset of logically related objects of the file system. In response to a migration request that requests migration of a selected logical volume from a source partition server to the target partition server, an ownership map is updated to indicate the selected logical volume is owned by the target partition server. At the source partition server, forwarding information is stored to reference the target partition server. When the source partition server receives a request for access to the object, the forwarding information is transmitted to the requester.
Various example embodiments are set forth in the Detailed Description and claims which follow.
REFERENCES:
patent: 5608903 (1997-03-01), Prasad et al.
patent: 5758344 (1998-05-01), Prasad et al.
patent: 5956718 (1999-09-01), Prasad et al.
patent: 6047289 (2000-04-01), Thorne et al.
patent: 6438652 (2002-08-01), Jordan et al.
patent: 6687701 (2004-02-01), Karamanolis et al.
Karamanolis Christos
Mahalingam Mallik
Muntz Daniel A.
Zhang Zheng
Hewlett--Packard Development Company, L.P.
Mizrahi Diane D.
Mofiz Apu M
LandOfFree
Logical volume-level migration in a partition-based... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Logical volume-level migration in a partition-based..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Logical volume-level migration in a partition-based... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3311194