My original vision was that a good tape based HSM system (Tivoli,…) and a tape robot would be able to do both the Point-in-time backups (to be sent offsite) and keep dual copies of live archived data, in one big solution.

However, This solution became more complex since my CIO wanted to include MAID technology as a secondary storage option. Since I think sending MAID disk offsite is a problem, I would need to pursue the offsite backups separately.

The the solution became more complex, because one of our largest servers (an image server) is a Apple XServe, and there seem to be few vendors that provide HSM agents for the Xserve. Several vendors proposed the we use a hybrid of NFS and some proxy agent. However, they did not seem happy with the solution. All the vendors we talked to (except one) said that we would save money in the long run, if we got off the XServe now ..

This is a politicol issue, since one of our departments has a strong loyalty to the XServe platform, and it has a good two years of life left in it … Then I’m confused about Apples statement that they are getting out of the XServe Raid business (http://www.macworld.com/article/132164/2008/02/raid.html)

Any ideas on how this should be approached

We would like to put our DFSMS/HSM ML1 data on FATA-disk. After some calculations we found out that a 150 TB second tier buffer would be big enough to store one year of ML1 data. Older data will be sent to a virtual tape system (SUN/STK VSM).

What are people’s opinions as to whether this is a good thing to do?