SQL backup very quick by snapshot

  • Hello,

    does have somebody experience with Netapp ? there is some kind of snapshot for for example one db , looks like frozen for few seconds .

    For example backup of big database with OLA backup procedures was about 20 hours, divided to more files is about 4hours . With Netapp is about 20second with snapshot.

    I never seen before and is some little different than I used normally. Do you have some experience with this ?

    http://docs.netapp.com/ocsc-41/index.jsp?topic=%2Fcom.netapp.doc.ocsc-dpg-mssql%2FGUID-10B0ECF9-A3DE-4B98-8D3E-6900F37B74CC.html

    Thanks

    • This topic was modified 5 years, 5 months ago by  tony28.
  • Yeah, you are talking about snapshots with Netapp's Snapmanager.

    When you say its frozen your users can actually still make transactions, its not like your database froze in the sense that no one can make transactions, it stills running and working, your sql server log will say it froze and it took a backup.

    We usually take snapshots specifying copy-only backups, it takes only 10 minutes in our environment, we still mainly use SQL Server and Ola as our main restore option.

    You can also use snapmanger to make full backups and transaction logs, it doesn't support differentials (We believe it takes way less time than SQL Server but we haven't tested it that way).

    You can use those backups to clone to other environments (Temporarily) since the snapshots make the aggregate to grow.

    After going through the doc you sent, you are actually using Netapp's Snapcenter which is the latest version, we are still using Snapmanager (not supported), we've been planning on migrating.

    • This reply was modified 5 years, 5 months ago by  Alejandro Santana. Reason: Added the last paragraph
  • Hi,

    thanks for info.

    Only question, if you have something like this, why you still using Ola ? Because i am thinking why is not written like quick backup solution, for example I didnt read before on some SQL blog or forums that something like this exists and is more faster and etc.

    There can be some problem with this ? I mean like restoring, backup or something. Because I dont have experience with this.

    I checked that during backup are saved to disk some metadata about backups. but not sure how works deeply.

    Thanks

  • tony28 wrote:

    I checked that during backup are saved to disk some metadata about backups. but not sure how works deeply.

    Yes, the metadata belongs to the snapinfo, you must never touch them, they keep track of all the backups/snapshots made inside the server with the netstorage, i don't know if its a must but they recommend a partition for the snapinfo, in our production servers we usually have a partitions with the size of 2GB, which nobody have access to but the snapmanager service user.

    We use Ola to have two ways to have backups, we have the phsyical made with SQL Server and snapshots taken with the SAN, so we can have 2 different ways to recover.

    We haven't  swapped fully to snapmanager because we have not made tests with our VLDBs, i don't really know what the SAN does when you take a full backup + transaction logs or where they are being kept, I don't know how a restore would work, i know how a clone works and our 6+ TB databases are cloned within minutes, Snapmanager Snapshots are basically photos of how the LUNS were in a point in time, don't know how it would work to normally use Netapp's Snapmanager for backups.

    We are looking forward to test how it would work in production and how the recovery process would be or if we migrate to snapcenter which is way more easy to manage than snapmanager.

    Also, like i said earlier, if you are gonna be using snapshots for QA/Development environment, make sure to refresh them weekly or monthly, because the more transactions they make the more affected your production Aggregate will be, as the snapshot grows it will start consuming your aggregate space and your production environment uses the same aggregate to grow, which means the snapshot and production are using the same storage, if it fills completely your production and other environments will crash.

    Greetings.

     

Viewing 4 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply