Data Backup and Recovery

snapcreator config with multiple DBs

brauntvr2swiss
2,820 Views

Hi All

Today we have following Environment:

- A Oracle Cluster with 10 Oracle DB Ressourcegroups in activ/passiv mode (All DBs runs normally on one Clusternode).

- One snapcreator configfile  with all DBs defined and  run  again the Clusternode, where hold all the Oracle DB ressources.

- We have 3 snapcreator Schedules (daily,wekly,monthly) for this configfile.

- In the Protection Manager we have one Dataset.

Generally this Environment run smoothly :-), but we have following restrictions:

- When we move one DB Ressourcegroup from one Node to the other, we have to edit everytime both configfiles (Node1: Remove Volumes/DB, Node2: Add Volumes/DB)

- One PM Dataset per Clusternode with RessorcePools = Nearstore Volumes and snapvault are generated automatic. We have a long retention on the nearstore and so its horrible when

  a DB Ressourcegroup is moved from one Clusternode to the other or a DB is moved to another Server, because this gives new snapvault relationsships in another Volume (because its another Dataset with its own Nearstore Volumes).

Future Design we want to implement for flexibility reasons:

- The ability to move the Ressourcegroups separate without edit the configfile in snapcreator.

- One Dataset in Protection Manager per DB Ressourcegroup (10 DB instances = 10 Datasets).

For reach this flexibiliy we must have  one Configfile for every DB and then we run the configfile against the DB Ressourcegroup and not the Clusternode (Backup runs then without editing any config).

In this Case we have 10 Configfiles instead 1 configfile

Question about this implementation:

- Does this mean, that we need three Schedules for every configfile (10 configs with 3 Schedules per config = 30 Schedules). Or gives a solution, where I can start all the configs with one Schedule

- Can I start all the Schedules (fe 10 daily) to the same time, because for some DB instances the same snapcreator client is involved (because this instances runs on the same physical Node) ....

- Is someone out there with such a kind of configuration?

I hope you understand my explanation 🙂

TIA

Thomas

2 REPLIES 2

ktenzer
2,820 Views

Currently using SC scheduler you can only create scheduler for a config not multiple configs

If you have multiple configs running at same time and both hit same agent that isnt good. The agent will queue, it can only do one thing at a time in its default mode. You can on unix only start agent with --start-multithreaded-agent which allows for parallel processing but again only on unix.

If you have multiple DBs on same host and need to backup them up at same time why not just back them up together with same config? In config you can set ORACLE_DATABASES=db1:oracle;db2:oracle etc...so you can have multiple DBs in there. You would probably want to back things up together if DBs shared same volumes etc. If DBs dont share same volumes it means restore will be single file only, you will never want to recover volume, ever, so you lose this capability if you go to shared backup. If DBs dont share same oracle home you can set ORACLE_HOME_SID so ORACLE_HOME_DB1=/path/to/orahome/db1 ORACLE_HOME_DB2=/path/to/orahome/db2.

In addition SC offers different schedules, so under NTAP_SNAPSHOT_RETENTIONS you can have daily:4,weekly:5,monthly:6. They you can run same job or config at different times with different retentions.

Let me know if you have further questions?

Regards,

Keith

brauntvr2swiss
2,820 Views

Hi Keith

Currently using SC scheduler you can only create scheduler for a config not multiple configs

Oky...

If you have multiple configs running at same time and both hit same agent that isnt good. The agent will queue, it can only do one thing at a time in its default mode. You can on unix only start agent with --start-multithreaded-agent which allows for parallel processing but again only on unix.

Snapcreator agent run on a Solaris System. So we can run multiple Configs at same time with the --start-multithreaded-agent.

If you have multiple DBs on same host and need to backup them up at same time why not just back them up together with same config? In config you can set ORACLE_DATABASES=db1:oracle;db2:oracle etc...so you can have multiple DBs in there. You would probably want to back things up together if DBs shared same volumes etc. If DBs dont share same volumes it means restore will be single file only, you will never want to recover volume, ever, so you lose this capability if you go to shared backup. If DBs dont share same oracle home you can set ORACLE_HOME_SID so ORACLE_HOME_DB1=/path/to/orahome/db1 ORACLE_HOME_DB2=/path/to/orahome/db2.

At this time our snapcreator environment run the way you describe above. Our problem in the future is following. We have a Oracle Cluster and the DBs could be on both Nodes of the Cluster. When I have one Configfile with all DBs in(like we do it today) and a DB went moved from one Node to the other, I have to edit the Configfile, otherwise the Backup from this moved DB fails. When I contact the different DBs over the Cluster Ressource Adress, it doesn't matter on which Node the DB runs and the Backup runs without edit anything. The other nice behave of contacting every DB individual is that Protection Manager creates a Dataset from every DB and all snapvault relationships from one DB (redo1/redo2/bin/dbf) are on one Nearstore Volume (fully automated).

In addition SC offers different schedules, so under NTAP_SNAPSHOT_RETENTIONS you can have daily:4,weekly:5,monthly:6. They you can run same job or config at different times with different retentions.

We do it this way today..

regards and thanks

Thomas

Public