Advertisment

Data Storage in SANs

author-image
PCQ Bureau
New Update

To understand how a SAN works, we must first understand the basic structure of a file. A file has two elements in it: data and metadata. While data is the actual content that needs to be communicated, metadata is information about the file itself, such as the name of the owner of the file, permissions (read/write), physical file location, volumes and its time of creation. For instance, if you create a Word document, then the content in that document is data, and the filename, path, permissions, etc, are all metadata. Ordinary file systems such as FAT, NTFS, Ext2 and DFS store both data as well as metadata together in the same file. In a SAN file system, however, the metadata and data are stored in separate places. Let's understand how a SAN file system works. 

Advertisment
Direct

Hit!
Applies to:

Organizations interested in deploying a SAN

USP:

Clearly explains how files are stored in a SAN
Links:

www.ibm.com, www.snia.org 

When a client on a LAN needs to access data, he first queries the metadata server for it. The metadata server then tells the client where this data is on the SAN. The client then accesses the storage device to get the data. This way, metadata processing and the actual data transfer happens separately, thereby speeding up the overall access. The storage devices on the SAN are connected to the metadata server through fiber-channel links. The clients on the LAN are also connected to the SAN through fiber-channel links, which enable faster, large-volume data transfers.

Let's compare the SAN file system to a normal file system. In an ordinary file system, files are organized in directories and sub-directories to form a tree-like structure. So the path to a particular file starts from the root directory, and moves up the directories till it reaches the sub-directory in which it's located. A SAN file system doesn't have this elaborate hierarchical directory structure. It has a single root fileset, and under this are all the other filesets. Each fileset contains a certain type of data, for instance-a Unix fileset with Unix data, a Win fileset with Windows files and an HR fileset with data from the HR application. The SAN administrator can define a storage limit for each fileset (quota). The filesets are further mapped to storage pools. 

Moving further, there are two kinds of storage pools: user and system. The system pool is a combination of the system and file metadata. Metadata servers use this system pool to locate files on the SAN. There can only be one system pool, but you can create multiple user pools. Each user pool can then point to a particular fileset. For placing data in user pools, policies are created. A policy contains rules that help identify which fileset should the data finally go to or be retrieved from. There's a default user pool as well, so if the file system can't identify the user pool to which data belongs, it automatically goes to the default user pool. 

Sushil Oswal

Advertisment