Memory & Storage
Engage with Intel professionals on Memory & Storage topics
21 Discussions

Sizing SAP workloads with Intel Optane Persistent Memory

Krishna_Yalamanchi
1 0 2,515

Simple sizing with PMEM

Your in-house SAP Basis Administrators or your Systems Integrator(SI) is able to answer these 2 questions for you

  1. How big is your SAP HANA database now(in GB or TB) now - Stored in PMEM
  2. What is the Working Memory / Scratch Pad size now? - Stored in DRAM
  3. 3 year growth in terms of # of users or DB size  - Growth Factor
    Eg. 10% is 0.10, 25% is 0.25, 50% is 0.50

Use the formula below to figure out the ratio of DRAM: PMEM
DRAM : PMEM  = (DRAM + DRAM * Growth Factor) : (PMEM + PMEM * Growth Factor)

 

Sizing Suite on HANA System

SAP Note  2786237 has 2 SQL files attached and run this script in your production system and they usually take < 5 seconds to run on your SQL

Do not run the SQL script immediately after a HANA downtime which does not capture the peak DRAM metric properly.

 

As a rule of thumb most SAP HANA transactional systems usually fit in the 1:2(DRAM:PMEM) ratio & HANA Analytical systems like SAP BW usually fit in a 1:1(DRAM:PMEM) ratio.

For Non Production systems like Dev / Sandbox / Training higher ratios of 1:4 are possible.

 

BW ON HANA Sizing Deployment with PMEM

Below is the sample output of the SQL  which we had run for a Customer running BW on HANA on a 4 + 1 scale out system today.

 

 

Krishna_Yalamanchi_0-1622072972726.png

 

 

Analyzing SQL Output

 

The above script indicates that DRAM:PMEM ratio is 1:2 and that the node size can go up from 1 TB(current system) to 9 TB / Node. In this case the customer migrated from Scale Out to Scale Up architecture for their SAP BW on HANA landscape.

 

Benefits with PMEM

Lesser IT Headcount to Manage / Patch and Upgrade 6 Scale Up Nodes vs. 15 Scale Out Nodes. Able to encrypt data in memory using native 256 AES encryption PMEM DIMM Encryption

 

Krishna_Yalamanchi_1-1622072972728.png

 

 

For an accurate sizing provide the below information to your OEM / Cloud Service Provider

 

  1. # of Environments Eg. Dev / QA / Production / Training
  2. Enterprise Storage Refresh strategy
  3. RTO / RPO requirements
  4. Planned Upgrades for next year / current year
  5. Non Production systems refresh strategy
About the Author
Krishna Yalamanchi is a Technical Solutions Specialist at Intel in the Memory Solutions Product Group. He focuses on enabling Intel Optane persistent memory solutions with Hyperscalers, Global System Integrators (GSIs) and OEMs.