TSM and NAS filers

[kalkrish]

Hello -
Below is an output from TSM for Windows B/A Users guide

A variation on the above approach is to use a Tivoli Storage Manager backup-archive client running on the NAS device, provided that the NAS operating system allows external programs. This method avoids the overhead of CIFS or NFS. Data can be stored on the Tivoli Storage Manager server with file-level granularity using progressive-incremental backup. The data is stored in the Tivoli Storage Manager storage hierarchy and can be migrated, reclaimed, and backed up to a copy storage pool. However, this approach requires data flow through the Tivoli Storage Manager client. This method also requires data flow over a network and through the Tivoli Storage Manager server unless a LAN-free configuration is used.

Is there any documentation on what NAS OS's accept the installation of external programs? or is this for something like openfiler?
Specifically is this available for netapp filers? if not is there any documentation that shows that it is not supported.
Thanks in advance

 

[KB]

NAS Backup

You have to consider the NAS operating system and file system. Is the operating system and filesystem supported by a TSM B/A client? NetApp filers have an internal operating system that TSM does not have a B/A client for. So you don't get the benefits of the TSM B/A client being able to deal with the file system directly, so no direct incremental backups, archives, or HSM. You just get the functionality that NDMP affords.

When you look at something like the Linux based openfiler, it opens up some options. You have to consider that this particular openfiler Linux distribution is not supported by IBM but as with other unsupported Linux distributions there ways to get TSM installed and working.

But if you are looking for a scaleable NAS system that is fully supported by IBM and has TSM functionality built into the product you should consider the new Scale out File System (SoFS)

http://www-935.ibm.com/services/us/its/html/sofs-landing.html

This NAS system is built on a Linux platform and uses IBM's General Parallel File System (GPFS) as its underlying file system. The GPFS file system has many desirable features such as being able handle billions of files and exabytes worth of data, snapshots, replication, file placement policy, etc. etc. GPFS is a TSM supported Unix file system so functions like backup, archive and HSM are there. The SoFS system is based on a scaleable cluster so I/O performance grows as the fully active nodes are added. The cluster capability is also what makes the backup process very interesting. GPFS takes care of figuring out what files have changed and gets them ready for an incremental backup. Multiple cluster nodes can take part in the backup process to get the I/O rates up in order to get the job done. Scanning the file system containing billions of files is fast. How fast? Obviously it depends on a number of factors but let's just say you will be measuring minutes and seconds rather than days and hours. Information lifecycle management (ILM) is built in so you can integrate disk type to disk type to tape file management (or any media that TSM supports) in the SoFS system.

What I like about SoFS is that it scales. You can start with a small system with entry level disk and then grow and grow, in both size and geography. The global namespace doesn't change and the file data blocks are moved around transparently to accommodate the new disk being added to the system. So you don