• Your sophisticated method for tape sounds like something similar to what we have. (NetBackup)

    I totally would like to write to disk and have them back that up to what ever medium they wish. This method would cost a good bit in storage but I think it is well worth the cost for stability and I can manage the backups for SQL. We have a happy mix of large and small databases majority are ranging from 1 - 10 GB and a handful of them upper 30 GB and 1-2 in the 300 GB range.

    As far as the relationship with the Network Admins it is not that it is bad, I guess its just neither knows enough of the other world :). I have been digging into the SAN information (NetApp) to understand it better and try and speak intelligently about the products but without being able to touch and play it is very difficult.

    After a little further research, the SAN uses a client end to interface with SQL much like the tape method you mentioned. The Client pauses, takes snapshot of SQL and then I think it triggers a drive level copy (not 100% sure how that fully works, yet). This side works and I use this on occasion to make copies for dev with the larger databases. Now keep in mind I use the client side to control how often they occur and type of backup. Currently, we kind of use this as a nice to have backup where tape is the major line of defense. As I mentioned, if the SAN backup method runs out of space then I have to clear some or all backups or I lose the drive. If I was in full recovery mode and taking log backups say every 15min and have to clear backups I lose the integrity of full recovery mode. with the added question what do I do for the tape backup if log backups occurring every 15min from another system (SAN). I would imagine a possible scheduling conflict could occur let alone a pain to restore. I haven't seen anyone mention how they balance the different methods and I can not make a recommendation to our team.

    The SAN we are using could be used as a source for tape backups, which I asked about. The reply is "We just don't have resources to dedicate to figuring out how as it requires lots of custom scripting". Anyway, perhaps someday we will find a better solution :).