Math's Network Appliance Server
The FAS 250 Network Appliance represents a network storage solution which solved several data management issues for our department. It provides us with centralization for our user files (which were previously dispersed among numerous servers and several operating systems) as well as a simplified backup and recovery strategy.
Initially funded by the 2004-05 ACIC, it was deployed for graduate student use for the Spring semester. At this writing, with additional disks and software purchased with departmental funds, the data of all classes of users have been successfully relocated to this network device. Still only 50% populated with disks (it holds 14), it currently gives us about 500gb of usable space.
Backups and Restores
Partially due to the legislation and the myriad legal requirements of data retention (see Chapter 404i), but primarily to the fact that our longtime backup/restore operations proved to quite inefficient when restoring large datasets, we needed a solution that was quick and reliable. Hence, the NetApp and the concept of snapshots (see details below). It also made good common sense that we follow the trend that the University central systems and particularly LCSR are using in terms of the network appliance devices. Since Math's servers are still located in the machine room in CoRE and since Math still relies on the student operators for the restoration function, we have a common strategy for disaster recovery as LCSR. Although the operational method for backing up the data has changed, the method for requesting a restoration has not. Users still make requests via e-mail to help@math.
Snapshots and Schedules
Snapshots are the backups that are performed at set intervals on the NetApp. They are similar to photographs in that they take an image at a given point in time. What essentially happens is that user data is not really deleted when files are changed or accidentally "deleted", the information (or pointer) to that file is what changes. So, a restoration is really just changing the pointer back to what it was at a given point in time. Snapshots do take up space although it is certainly rather small compared to the data itself. Here is the current schedule of snapshots for our NetApp:
- Hourlies - taken every 4 hours and kept online for 6 days
- Dailies - taken daily (except Sunday) and retained for 2 weeks
- Monthlies - taken once weekly (Sunday) and retained for 9 weeks
We still use the older method of backing up to tape for off-site data retention. This is still in the process of being modified to more accurately reflect our new needs.