Plug-ins for Enterprise Applications(SQL & Oracle)

[Hirosh]

Hi guys,

we are planning to backup databases of our customers which reside inside VMs.im curious if we use the sql enterprise plug-in, how can we backup the database to have the minimum RTO and a RPO of 0? i have something like journaling in my mind, but i don't know how exactly can this be implemented? i don't want to use mirroring nor cloning, i just want to implement this with Veeam .im open to new approaches as well.


thanks in advance.

[Andreas Neufert]

You have 3 option:
Enterprise Plug-ins: You can achive a RPO of roughly 5 min, as you need to complete the previous log backup before the next one starts. OS is not covered directly (need to backup separately). RTO depend on the restore scenario and you need at least the time to recover the data (you need to transport the data back). It is the most flexible and in the database processing (like Oracle RMAN) integrated solution.

VM Image level backup. My guess is that you use VMware as Hypervisor. You can enable Guest processing with transaction log backups. RPO is rougly 5 min (you need to complete the previous log backup before the next one starts). OS is covered in the same way. RTO depends on the restore scenario. If you want to have the restore point of the image level backup (maybe daily/hourly) then it is ~2min + OS boot time for Instant Recovery. If you restore on the database level then this depends on the amount of data you want to recover (time for transporting the data).

For VMware there is Veeam CDP that replicates all IO from the VM to another system. RPO is "seconds" to just replicate the data. RTO is similar to instant recovery to any point in time. From a DB perspective the replication is unknown and you replicate the database in a crash consistent state, however you can set for Windows VSS consistent regular states so that you have from now and then a clean state that you can fallback if recovery from crash consistent state does not bring up the DB.

I would like to mention that in many cases people forget that to the RTO time as well Veeam external factors count. How long does it take to detect that a service is down and that you then decide that it needs a recovery and actually can start the recovery? Is the hardware down and you need replacement? Do you need other departments or tickets that you need to start the recovery and what is their reaction and resolution time?

To have a RTO/RPO time of 0 you need a database integrated HA system that failover automatically with a fast restore method in case the HA cluster fails as well. If you only want to go with Veeam, then I suggest a combination of Veeam CDP and Image level VM backup with transaction log backups.

[Hirosh]

@ Andreas Neufert

thank you for your thorough explanation, i will need to go through this to see which approach is more suitable for us.