Db2

 View Only

 Capacity to be planned- HADR

subhendu mohanty's profile image
subhendu mohanty posted Tue December 02, 2025 04:23 AM

Hello,

   We are hosting a solution on public cloud with DB2 database configured as HADR. The primary region will have 2 nodes of DB2 setup as active passive configuration with the secondary node replicated as SYNC. The DR region will have a single node configured as ASYNC. If application needs 32 VCPU , we are taking reserved instance as 32 for primary, 32 vcpu box for secondary and another 32 vcpu for DR database.My question is do I need to take a capacity of 32 core for DR or I can use a 8 core type of VM as there is no read/write traffic coming to DR and its primary objective is to stay close to data replication from primary.  We can always scale up in case of DR happening. Please guide.

Damir Wilder's profile image
Damir Wilder IBM Champion posted Wed December 03, 2025 04:13 AM

Hi,

You can certainly start with an 8-core DR node (I would if I were you), especially so if scaling up is not an issue, and see where it takes you.
As you said, if the DR node's only (regular) purpose is catching up with the primary node, then stacking up 32 cores just for that purpose looks a bit over the top.
It is certainly more important to have faster disks on the DR node, as that is what will be heavily used there (when replaying the logs), rather than CPUs.

Hope this helps...

Regards, Damir

Paul J Pearon Jr's profile image
Paul J Pearon Jr posted Thu December 04, 2025 09:56 AM

I will add that on the full capacity system, I usually review the Replay Workers startup message in the db2diag.log.  Then I base my initial reduced capacity system cores estimate on this.

Related Content