Automatically redistributing data of multiple file systems in a distributed storage system

Patent metadata

Field	Value
Publication number	US-9811529-B1
Application number	US-201313760933-A
Country	US
Kind code	B1
Filing date	Feb 6, 2013
Priority date	Feb 6, 2013
Publication date	Nov 7, 2017
Grant date	Nov 7, 2017

Abstract

A distributed storage system maintains multiple logically independent file systems. Each file system includes a data set stored by a distributed storage of the distributed storage system. During operation, access pattern levels for the multiple logically independent file systems are determined. Thereafter, the data sets included in the multiple logically independent file systems are redistributed across multiple storage devices of the distributed storage. In one aspect, redistribution of a particular data set is based at least in part on the particular file system including the particular data set and on the determined access pattern levels for the multiple logically independent file systems. In one implementation, redistribution is performed according to a uniform redistribution scheme. In another implementation, redistribution is performed according to a proportional distribution scheme.

First claim

What is claimed is:

1. A computer-implemented method for redistributing data stored in a distributed storage, the method comprising:
maintaining a plurality of logically independent file systems, wherein each file system includes a data set stored by the distributed storage and metadata including a unique identifier and organizational structure information;
observing accesses to data files in the data set included in each of the file systems to determine access pattern levels;
determining a respective access pattern level for each of the plurality of logically independent file systems;
determining that a first file system from the plurality of logically independent file systems has an access pattern level specifying a higher probability of future access than a probability of future access specified by an access pattern level of a second file system from the plurality of logically independent file systems;
determining storage requirements for a plurality of file systems;
obtaining device characteristic information for the plurality of storage devices of the distributed storage;
redistributing the data set of the first file system having the higher probability of future access before redistributing the data set of the second file system across a plurality of storage devices of the distributed storage; and
redistributing data sets across the plurality of storage devices based on the storage requirements for the plurality of file systems and the obtained device characteristic information for the plurality of storage devices of the distributed storage;
wherein redistributing the data sets across the plurality of storage devices based on the storage requirements for the plurality of file systems and the obtained device characteristic information for the plurality of storage devices of the distributed storage comprises:
determining a respective performance level for each of the plurality of storage devices based on the device characteristic information;
computing an aggregate performance level for the distributed storage by summing the determined respective performance levels for the plurality of storage devices of the distributed storage;
computing a proportional performance level for a particular storage device of the plurality of storage devices by dividing the determined performance level for the particular storage device by the aggregate performance level;
computing a target amount of storage space assigned to the first file system from the plurality of logically independent file systems for the particular storage device by multiplying a determined storage requirement for the first file system and the proportional performance level for the particular storage device; and
redistributing the data set of the first file system based at least in part on the computed target amount of storage space assigned to the first file system for the particular storage device.

2. The computer-implemented method of claim 1 , wherein
redistributing the data set of the first file system comprises redistributing the data set of the first file system uniformly across the plurality of storage devices of the distributed storage.

3. The computer-implemented method of claim 1 , wherein redistributing the data sets across the plurality of storage devices based on the storage requirements for the plurality of file systems and the obtained device characteristic information for the plurality of storage devices of the distributed storage comprises: redistributing the data set of the first file system from the plurality of logically independent file systems across the plurality of storage devices in proportion to device characteristic information.

4. The computer-implemented method of claim 1 , wherein redistributing the data sets across the plurality of storage devices based on the storage requirements for the plurality of file systems and the obtained device characteristic information for the plurality of storage devices of the distributed storage comprises: calculating a target amount of storage space assigned to a first file system for each of the plurality of storage devices by dividing a storage requirement for the first file system by a number of the plurality of storage devices; and redistributing the data set of the first file system across the plurality of storage devices based on the calculated target amounts of storage space.

5. The method of claim 1 , wherein:
the data set of the first file system and the data set of the second file system each comprises a plurality of data files.

6. The method of claim 1 , wherein:
the first file system and the second file system are each associated with a respective type of data selected from the list of: work data, temporary data, log data, and backup data; and
the first file system and the second file system are not associated with the same type of data.

7. A non-transitory computer readable storage medium executing computer program instructions for redistributing data stored in a distributed storage, the computer program instructions comprising instructions for:
maintaining a plurality of logically independent file systems, wherein each file system includes a data set stored by the distributed storage and metadata including a unique identifier and organizational structure information;
observing accesses to data files in the data set included in each of the file systems to determine access pattern levels; determining a respective access pattern level for each of the plurality of logically independent file systems;
determining that a first file system from the plurality of logically independent file systems has an access pattern level specifying a higher probability of future access than a probability of future access specified by an access pattern level of a second file system from the plurality of logically independent file systems;
determining storage requirements for a plurality of file systems;
obtaining device characteristic information for the plurality of storage devices of the distributed storage;
redistributing the data set of the first file system having the higher probability of future access before redistributing the data set of the second file system across a plurality of storage devices of the distributed storage; and
redistributing data sets across the plurality of storage devices based on the storage requirements for the plurality of file systems and the obtained device characteristic information for the plurality of storage devices of the distributed storage;
wherein redistributing the data sets across the plurality of storage devices based on the storage requirements for the plurality of file systems and the obtained device characteristic information for the plurality of storage devices of the distributed storage comprises:
determining a respective performance level for each of the plurality of storage devices based on the device characteristic information;
computing an aggregate performance level for the distributed storage by summing the determined respective performance levels for the plurality of storage devices of the distributed storage;
computing a proportional performance level for a particular storage device of the plurality of storage devices by dividing the determined performance level for the particular storage device by the aggregate performance level;
computing a target amount of storage space assigned to the first file system from the plurality of logically independent file systems for the particular storage device by multiplying a determined storage requirement for the first file system and the proportional performance level for the particular storage device; and
redistributing the data set of the first file system based at least in part on the computed target amount of storage space assigned to the

Assignees

• Quantcast Corp

Inventors

• Rus Silvius V
• Subramaniam Thileepan

Classifications

• G06F3/067

  Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS] · CPC title

• G06F17/30584

  Physics · mapped topic

• G06F17/30194 Primary

  Physics · mapped topic

• G06F3/0613

  in relation to throughput · CPC title

• G06F3/0689

  Disk arrays, e.g. RAID, JBOD · CPC title