System replication in HA and DR scenario [enquiry]

[former_member146669]

Dear Floks,

In a landscape with 3 HANA appliances where 2 for HA (production site) and 1 for DR (remote site), system replication is take place.

Both HA and DR will apply system replication (we don't use storage replication and backups method).

Between HA nodes:                     synchronous mode

Betwwen HA and DR nodes:        asynchronous mode

And per my understanding, there are primary (A) and secondary (B) system role in the production. Log are syncrhonize from A to B.

So, how about the system (C) in DR ?  Is the replication from A to C or B to C here ?

What's/Any dependence of the replication between A and B vs  A/B and C?

Many thanks.

Regards

Gary

[tamil_arasan · Accepted solution]

Hi Gary,

I never came across this scenario however as per below link, lets assume site A, site B and site C. Site A is primary for site B and site B is primary for site C. please check the link here

Using command line:

SAP HANA Administration Guide - SAP Library

Using studio :

SAP HANA Administration Guide - SAP Library 

Hope this helps,

Regards,

Pradeep

[former_member146669 · Accepted solution]

Thank you Pradeep.

I read the link and it is exactly what I'm asking.

[tamil_arasan]

Hi Gary,

Yes, you're correct. I hope in that case, we need to manually take care of the replication setup configuration. for example; let's assume site B is crashed, we need to manually re-setup the configuration of replication.  A site is primary and C site is secondary until the B site comes back online.

(A-->B-->C) is normal replication scenario

(A-->C) is replication scenario when site B is unavailable or crashed (Need manual system replication setup)

Since downtime is not required in the primary server for re-setup of replication with site C, system will be available to the users.

Thanks,

Pradeep

[former_member146669]

By the way, according to this setup, remote DR replication will be impacted if HA synchronization in PRD site fail or Node 2 have problem.

The advantage I can think of is to offload the Primary system resouce utilization (3rd node depends on 2nd node only and have no direct contact to 1st node).