cancel
Showing results for 
Search instead for 
Did you mean: 

Global Allocation Limit on Multinode

Former Member
0 Kudos

All

we have four nodes in our production environment with each 512 GB. Hana 1 is Master, Hana 2 and Hana 3 are defined as slaves. Hana4 is configured as Standby mode and its passive at this point of time.

We need to set Global Allocation limit of 512 GB on this box. Do you know SAP recommendation on setting GAL on multi node environment. Do we need to consider setting GAL for Standby node as well. Current memory usage on the Hana nodes.

HANA 1 # Master --> 98 GB

HANA 2 #  Slave --> 175 GB

HANA 3 #  Slave --> 253 GB

So whats the best and recommended way of configuring GAL in these scenarios. If we configure GAL on standby , then we are going to lose that Licenses memory as its not being used. Its used only when master node fails.

Mahesh Shetty

Accepted Solutions (0)

Answers (1)

Answers (1)

Former Member
0 Kudos

Hi Mahesh,

Based on our interaction with the Hardware vendor and SAP during the setup of a 4 Node Scale out HANA system(3 active+ 1 standby) the GAL was set for the entire SYSTEM instead on HOST level

The option to configure GAL on HOST level is useful when we have more than 1 HANA instance on the same appliance or for some reason if one of your nodes has to use less memory on a particular host

In our case we have 1 HANA instance, 4 HOSTS with 512 GB of RAM and our license is for1.5 TB

So we have set GAL to 0 on SYSTEM level; HANA can use the complete available memory

Please note as standby node does not have an active IndexServer running the GAL setting does not take into account the standby node

Setting the GAL on SYSTEM level helps in keeping the configuration simple

Hope this answers your question

I do have a few points that may need clarification below, so I can provider more inputs

What is the licensed memory of your HANA system? Do you have multiple instances of HANA on the same hardware? Is your standby node used as an active node for another instance?

Sunil

Former Member
0 Kudos

Sunil

We are licensed for 512 GB only. So we cannot set 0 at system level in our scenario and hence we need to restrict GAL limit on the node level.  We don't have another instance running on the same hardware and also standby node is not used anywhere else. I am attaching the system landscape FYI

Mahesh Shetty

Former Member
0 Kudos

Hi Mahesh,

Thank you for sharing the setup

You may consider using the option of GAL=170 GB on all 4 hosts. Effective memory usage would then match the license (approx 510 GB), this way you do not have to make any changes when the standby node has to take over the role from any of the failed nodes

Make sure the table redistribution is in place to make sure the tables are evenly distributed across the nodes

I am assuming this is a single container system and not MDC, can you confirm the same?

Sunil