We are using NetWorker Module for Oracle to back up a database with 1 TB data size. My Database team needs the RTO to be 1 hour. At the same time they need that we should be able to recover the data upto four days back.
When we recover from NetWorker complete database it takes about 12 hours and if we recover a single tablespace it takes around 2 hours.
I asked them to replicate the data but they told that in case of replication they can not go back to three or four days.
Any hint, what should we use in our setup to achive this.
It sounds like your DBA team wants to eat their cake and cookie too. they want to be able to restore from 3 or 4 days ago but they don't want to allocate enough storage to do so. Like I tell my management this isn't magic. You cant continually add clients or increase the amount of data being backed up without providing more CALS or tapes (and or disks) to the Backup Admins and still expect for backups to be performed adequately. You also cant change the laws of physics. If it takes 12 hours to backup 1 TB of data it will take roughly the same amount of time (more for tape loads, unloads and tape repositioning) to recover it. If they want to shrink the RTO than they need to shrink the amount of time it takes for backup.
I guess that restore should be faster, but this really depends on your network setup and disk layout/setup and tuning of course. Even when going to use PowerSnap, you need to make some smart decisions on disk usage and their layout.
For start, you would need to identify why this speed is such as it is. Is it network? Is it disk? Is it application? You would need to test all 3 of those to isolate this and see where bottleneck is and what can be done to improve it. That would be start. Without knowing what really limits you it is hard to take any further steps or plan anything further.
What will be those parameters as after few hrs I am having a meeting with my managers, DBA team and storage teams. There this information will help me a lot.
dgreggs1
86 Posts
614
0
Posted November 20th, 2009 06:00
It sounds like your DBA team wants to eat their cake and cookie too. they want to be able to restore from 3 or 4 days ago but they don't want to allocate enough storage to do so. Like I tell my management this isn't magic. You cant continually add clients or increase the amount of data being backed up without providing more CALS or tapes (and or disks) to the Backup Admins and still expect for backups to be performed adequately. You also cant change the laws of physics. If it takes 12 hours to backup 1 TB of data it will take roughly the same amount of time (more for tape loads, unloads and tape repositioning) to recover it. If they want to shrink the RTO than they need to shrink the amount of time it takes for backup.
Suggestions:
Oracle RAC
Oracle Data Guard
Third Party Replication