cancel
Showing results for 
Search instead for 
Did you mean: 

Disaster Recovery for HANA

Former Member
0 Kudos

Anyone knows the Disaster Recovery solution for HANA? Thanks,

Accepted Solutions (0)

Answers (2)

Answers (2)

Former Member
0 Kudos

Exactly, the attached doc regarding HANA standby means for high availibility not disaster recovery. if the entire data center goes belly up, there will be no shared disk available. I talked with a few folks from SAP BI area. It seems to me that at this point the best way to handle HANA disaster recovery is manual:

1. Back up HANA to local backup system (storage or tapes) at regular interval (e.g. nightly) and ship (ftp/ship etc.) the back up to DR site.

2. Back up HANA log more frequently (e.g. hourly) and ship the log backups to DR site.

In the event of a DR, the backups shipped to the DR site will be used to restore a HANA DR system. Of course, nomal DR practice still applies, such as, use of virtual names everywhere, change DNS entries, etc. to make the resumption of DR processing at DR site seamless ...

This is very similar to the very early days of RDBMS DR practice. I expect SAP will make the HANA DR less painful by implementing something similar to modern days RDBMS DR solution (e.g. database mirroring, standby db, replication, etc.) in the future.

Former Member
0 Kudos

HANA is not yet 100% matured for OLTP applications, so that SAP targets for OLAP applications, at the moment for DR, take care by OS tools

Thanks & Regards

Rao

Vitaliy-R
Developer Advocate
Developer Advocate
0 Kudos

Lu, you can read about B&R and HA at help.sap.com/hana. Any more details what specificaly you are looking for?

Former Member
0 Kudos

Vitaliy, actually I still don't understand this. There is this document: http://help.sap.com/hana/hana1_imdb_scale_en.pdf which is very vague. It alludes to the fact that failover only works when all of the data is stored on a shared disk. What happens if you have one system in Vienna and one in Milton Keynes and the Vienna office explodes, how on earth is it going to failover to the other server?