The official blog for Windows Server Essentials and Small Business Server support and product group communications.
EPS Team Blogs
I have heard a lot of talk about verifying a backup. Could you cover "best practice" on verifying a backup.
1) If you index the file and are sucessful, is this good enough? (my answer is NO)
2) If you are able to restore a test group of files back to the server, should you expect that you can recover all the files in your backup.
Thanks for your podcasts - I haven't deleted a single one on my MP3. 30Gb will hold a lot of podcasts. I'm an enterprise guy that has a small practice consulting for Small Business Server network installations. I totally GET the wizards (avoiding the Susan Bradley 2x4) and wouldn't have it any other way.
That's an excellent question - You just gave us a topic for next week. Let me try to cover the highlights:
>1) If you index the file and are sucessful, is this good enough? (my answer is NO)
Agreed. Indexing only tells me what file are on tape. You can index 100% of your files, and 100% of them can be corrupt (been there, done that). Likewise, indexing tells you nothing about the state of the physical tape (or hard drive0, or the header information.
>2) If you are able to restore a test group of files back to the server, should you expect that you can recover all the files in your backup.
That's usually our first test. If you can restore 3-5 random files, you can have a high degree of certainty that the backup is good. In support, this is our test before we go forward with any write operation on the server. This isn't a 100% certainty, though, since any individual file can have a problem (I've been bitten by this twice in 6 years - both times with DAT tapes that had been rotated for *years*).
If you want complete certainty that a backup is good, you really have to do a practice disaster recovery. This is a great offering to sell to your clients - once or twice a year go through a complete, bare-metal restore of the OS and the backup. This gives you certainty that your backups are working properly, and reduces your recovery time in a real emergency because you already have everything in place when the worst happens. I know of several law offices and securities firms that see the benefit of paying for this, but it can be a hard sell to many SMBs.
Susan has some good points:
I hope that above isn't interpreted as "blow away the production server and hope for the best on restore". Don't do that. In perfect consulting practice (tm) you'd standardize on hardware and bring in a nice, shiny replacement box for proof of concept. If you can't do that, we've seen very successful consulting practices sell the DR package as an offering done on the weekends or after hours, where the server is left in place, but they shut down the server, remove the hard drives (carefully) and slap in a new piece of bare metal to do the restore. Since the HD is really the only thing besides the fan in your server that has moving parts, it's the only thing likely to break (and, boy, do they break...every day). So, replacing the HD and restoring to a base server install is really the most real world scenario you can prepare for.