01-09-2017 05:39 AM - edited 01-09-2017 06:08 AM
Hi, we use NetBackup 7.7.3 with a DataDomain 990 and try to start with SQL backups which are now backed up as dump to disk.
Our DBA has demanded transaction log backups every 10-15 minutes for a few db's and every 1 hour for the rest.
I now have 4 testservers configured in an Intelligent Policy making whole_database backups. It totals to 38 db's.
The transaction log backup of these test-db's (without any modifications) takes 15 minutes, the SQL Server tab is all default except for 10 Parallel operations. Every job takes about 3 minutes to complete.
Extrapolation to 260 servers with 2500 db's with real modifications tells me that I never reach a transaction log schedule of 1 hour, let alone 15 minutes.
So I have several questions:
- Why does a single 32 KB job takes 3 minutes to complete.
- How to speed up this process.
- Can I ever backup 260 servers this way.
- How to exclude the 'standard db's' like 'model' etc. from transaction log backups (to lower the number of jobs).
01-09-2017 05:55 AM
Can't help to wonder if the backup system was sized for this intended load. See the performance tuning & planning guide.
Think you need to get together with the SQL DBA's if you havn't already, as you probably will need their help to identify the bottleneck.
Have you looked at the performance suggestion in the SQL admin guide ?
How exclude databases at least for scripts is also described in the SQL admin guide
How much of the 3 minutes is actually data transfer ? you might have skewed results as the meta data part is an abnormal big part of the test jobs.