Re: Two Disaster Recovery Questions
From: Geoff N. Hiten (SRDBA_at_Careerbuilder.com)
Date: 01/27/05
- Next message: Sierra: "Re: How to View simplified chinese !"
- Previous message: ChrisR: "Re: Update the value of @@ServerName"
- In reply to: JM: "Re: Two Disaster Recovery Questions"
- Messages sorted by: [ date ] [ thread ]
Date: Thu, 27 Jan 2005 11:23:31 -0500
Good for you. End-to-end testing is the only way to be sure your DR plan
really works. You are definitely ahead of a lot of places. Not here, but
that is another story.
-- Geoff N. Hiten Microsoft SQL Server MVP Senior Database Administrator Careerbuilder.com I support the Professional Association for SQL Server www.sqlpass.org "JM" <JM@nospam.com> wrote in message news:%23Jh%23kfIBFHA.2076@TK2MSFTNGP10.phx.gbl... > Thanks to all for your responses. Some answers to your questions: > > The hotsite isn't dedicated, so It's probably more appropriate to call it a > recovery site (Sunguard in Philly, PA). We did not have time to run a > second restore at the recovery site. > > Both servers are configured as RAID5 and consist of five 10k rpm disks. We > use Dell servers, the server at the recovery site was a Compaq this time > (the drives located inside the server chassis in both cases). I knew that > raid controllers varied in performance, but I didn't expect that big of a > difference. > > Detaching from the database is acceptable in our environment, our systems > are available from 6a-10p CST. I may have to look into this further. > > When we practice the recovery (both in our office and at the recovery site), > the servers are built up from scratch (install OS, apps, restore data, > configure, test, etc.). > > > > > "Geoff N. Hiten" <SRDBA@Careerbuilder.com> wrote in message > news:OFXtp$HBFHA.904@TK2MSFTNGP12.phx.gbl... > > The additional time is likely due to having to create and initialize the > > data files from scratch. If your test restore overwrites the existing > > databases, that can show a major speed improvement. Also, even if the dis > > configuration is 'similar', a difference in SCSI subsystem type can > account > > for a major speed decrease. > > > > -- > > Geoff N. Hiten > > Microsoft SQL Server MVP > > Senior Database Administrator > > Careerbuilder.com > > > > I support the Professional Association for SQL Server > > www.sqlpass.org > > > > "JM" <JM@nospam.com> wrote in message > > news:eC93wlHBFHA.2180@TK2MSFTNGP12.phx.gbl... > > > Question #1 > > > We run a full backup nightly that produces a 30GB BAK file. In disaster > > > recovery testing in our office, the SQL restore operation takes about an > > > hour, however, when we tested our procedures at the recovery "hotsite", > it > > > took 2.5 hours to restore the database. Why the big difference in time? > > > Granted, the processors in the hotsite machine were a little slower, but > > the > > > processors were almost completely idle. RAM was the same, and the hard > > disk > > > config was very similar, and other factors that I can think of were also > > > similar (BAK was on local disk, etc.) The restore rate was 7.496 > MB/sec. > > > > > > Question #2: > > > To speed up recovery, someone asked whether we can backup and restore > the > > > MDF/NDF/LDF files directly, thereby avoiding having to run the BAK's > every > > > night, and the pain of running the sql restore during a recovery. I > > assume > > > that the SQL service would have to be stopped during backups and > restores. > > > Is this possible? Is this acceptable? > > > > > > Thanks! > > > > > > > > > > > > > > >
- Next message: Sierra: "Re: How to View simplified chinese !"
- Previous message: ChrisR: "Re: Update the value of @@ServerName"
- In reply to: JM: "Re: Two Disaster Recovery Questions"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
