Trying to come up with viable backup restore & logshipping solution for achieving the following:
- 15 minutes Recovery Point Objective (no more than 15 minutes data loss at any time)
- 5 minutes Recovery Time Objective (must be able to get the db up and running back by 5 minutes)
Considering using logshipping only (which I think is kind of pushing it, but I want to know if anyone else know how to achieve this).
Some other info for consideration:
- Using 40 Gbit / sec fiber channel between the primary and disaster recovery (DRC) sites
- The sites are about 600 km apart.
- At close of business, the amount of data generated is predicted to be about 150 MB/sec.
- Log backup is planned for every 5 min.
Doing some rough calculation I came up w/ the following numbers:
- 40 Gbit / sec = 5 MB / sec @ 100% network efficiency.
- 5 MB / sec = 300 MB / min.
- @ 300 MB / min, the total amount of data that can be transfer considering the 5min RTO is about 1.5GB, but that will left no time for the actual backup and restore, so if we cut it down to 3min logshipping time, which equals to ~900 MB over 3 minutes at 100% network efficiency, that will left about 1 min backup time and 1 minute restore time. Currently don't have any information if the system being used is capable of restoring 900 MB in 1 min, but assume it can.
- for COB scenario... 150 MB/sec, and considering the 3 min logshipping time, which should equal to about 27 GB of data over 3 mins...??? I think this is where the SLA will break... since there is no way to transfer 27 GB of data over a 40Gbit/sec line in 3 min.
Can I get someone else opinion?
I am thinking database mirroring might be a better answer for this...