Distributed file data location
View Patent ↗Locating data within a heterogenous distributed file system is difficult due to the many file access standards in use. A file system for easily locating data includes storage devices holding at least one copy of each file. At least one location server maps a file identifier for each file into the location of a copy of the file represented by the file identifier. One or more name servers map a file name into the file identifier referenced by the file name.
1. A file system for storing data comprising:
a plurality of storage devices, each storage device operative to store at least one copy of at least one file;
at least one location database comprising a map between a file identifier for each file and location information for each copy of the file represented by the file identifier;
at least one name database comprising a map between a file name and the file identifier referenced by the file name, each name database physically separate from the at least one location database; and
at least one client operative to
(a) request a file identifier corresponding to a requested file name,
(b) receive the file identifier mapped to the requested file name,
(c) request location information corresponding to the received file identifier,
(d) receive location information mapped to the received file identifier, and
(e) access data using the location information.
2. A file system as in claim 1 wherein each file is stored as at least one file extent, the file identifier comprising a file handle.
3. A file system as in claim 1 wherein each file is represented in storage as an object and each file identifier is an object identifier.
4. A file system as in claim 1 wherein the client is further operative to access file metadata stored in the location database.
5. A file system as in claim 1 wherein the client is further operative to send the file identifier and a new file name to at least one name database, thereby registering the new name for the file.