Skip to Content
Product Information
Author's profile photo Stefan Baeuerle

Update – SAP HANA Cloud to Support ARM-based AWS Graviton Processors | Driving Price-Performance Improvements and Sustainability via Energy Efficiency

After several months of close collaboration and teamwork between the SAP and AWS teams, we have successfully achieved several milestones to support SAP HANA Cloud on ARM-based AWS Graviton processors. This is an ambitious joint effort between the SAP HANA Cloud and AWS engineering teams, including AWS’s consulting unit, AWS Professional Services. As we shared in our blog post last year, SAP and AWS have a longstanding partnership with thousands of joint customers. This effort to enable SAP HANA Cloud on AWS Graviton is a cooperative strategic investment and supports our customer’s goal to become an intelligent, sustainable, enterprise.

This project encompasses migrating several different cloud building block components, including common infrastructure services and managed Kubernetes clusters as well as the SAP HANA database container itself. This represents approximately 300 services required to be ported to the ARM-based AWS Graviton infrastructure. The diagram below shows the components in-scope within SAP HANA Cloud and the SAP Business Technology Platform.Infrastructure%20Layers%2C%20SAP%20HANA%20Cloud%20on%20ARM

We are pleased to share that the SAP team has successfully deployed an SAP HANA database container on AWS Graviton in SAP HANA Cloud as part of a first end-to-end proof point. In addition, all key milestones planned have come together, including the migration of Gardener managed Kubernetes Cluster, Build infrastructure, and the deployment of containerized microservices via the SAP Common Service Infrastructure onto AWS Graviton. This successful milestone included thousands of unit and integration tests to validate the end-to-end capabilities.

The initial price-performance assessments of running SAP HANA Cloud on ARM-based AWS Graviton processors are in-line with our planned expectations to reduce total-cost-of-ownership as well as lower power consumption. As previously noted in our first blog, Graviton3-based instances use up to 60% less energy for the same performance than comparable EC2 instances. In the upcoming weeks, the team will focus on additional performance optimizations and testing the first selected SAP HANA Cloud-based applications on the AWS Graviton infrastructure.

Today, hundreds of SAP and BTP applications and cloud services, including SAP Datasphere and SAP Analytics Cloud, as well as thousands of bespoke custom applications leverage SAP HANA Cloud as their database-as-a-service foundation. We anticipate the initial SAP HANA Cloud workloads on AWS Graviton infrastructure to be rolled-out in Q3 2023 in selected SAP HANA Cloud AWS regions and availability zones for specific applications and environments. It is our ambition to migrate the SAP HANA Cloud instances currently running on EC2-based AWS infrastructure to AWS Graviton, that will benefit from the improved efficiencies that Graviton is known for, such as price performance and reduction in overall carbon footprint.

There will remain one singular SAP HANA Cloud service, made available to deploy in the AWS Regions and Availability Zones of choice. The transition will be a seamless experience for our customers with the migrations performed during planned maintenance windows for all existing SAP HANA Cloud features and capabilities.

The development teams working on this initiative are proud and excited for what they have accomplished and more importantly, what this means for our joint customers in terms of price-performance, sustainability, and innovation. We would like to thank our colleagues at AWS for their continued support throughout this initiative. This close partnership with SAP and AWS is only the beginning and we are looking forward to continuing to co-innovate together in the future. Stay-tuned for our final update later this year once SAP HANA Cloud running on AWS Graviton becomes available.

Assigned Tags

      You must be Logged on to comment or reply to a post.
      Author's profile photo Patrick Rieken
      Patrick Rieken

      Hi Stefan,

      This is great news, congratulations !
      Free choice of CPU vendor, OS and before Hana DB,  DBMS was always key to SAP.

      This leaves me with the following question: what about AMD Epyc ?
      10 years after releasing Hana, AMD still is not supported for production use.
      Current Epyc generation meanwhile runs circles around Xeon w.r.t. Performance-Efficiency..

      I double-checked in the QA of the "whats new in Hana 2.0 SPS07 yesterday and the answer was:
      There are no plans to support AMD CPU generations with SAP HANA in the foreseeable future officially.. There are several reasons why this is not planned, from costs of porting SAP HANA, up to missing support and infrastructure by partner.

      • I guess porting to IBM Power & AWS Graviton ARM is of an much larger scope compared to AMD Epyc, which is already x64 compliant.
      • AMD not investing in providing needed support and infra to port ? And therewith missing a huge market segment ?

      Really hope Graviton starts shining for Hana, and very curious to the what about for AMD..


      Regards, Patrick Rieken.

      Author's profile photo Stefan Baeuerle
      Stefan Baeuerle
      Blog Post Author

      Hi Patrick,

      Thanks for the question. We need to distinguish the work that we do for SAP HANA Cloud and our releases for SAP HANA 2.0, like the recent release of SAP HANA 2.0 SPS 07.

      The work that we do on ARM-based AWS Graviton is focused on SAP HANA Cloud. While we are doing the porting, the specific machines that SAP HANA Cloud runs on will not be exposed. Like for other cloud services, consumers will not see on which exact machine or infrastructure a given instance runs. That would also be true for other options.

      And to be even more precise: Our ambitions are focused on optimizing SAP HANA Cloud as much as possible from a cost of ownership (TCO) and performance point of view (optimizing price-performance), based on the given conditions within the cloud service providers’ environments. Current focus for now is on ARM-based AWS Graviton. While other portings for SAP HANA Cloud might potentially come in the future, this is not planned in parallel to the ongoing porting efforts and optimizations.



      Author's profile photo Tom Dhooghe
      Tom Dhooghe

      Good to hear this initiative is on track Stefan! It does raise the question though of how this will affect our customers. Given the potential price-performance gains, once AWS HANA Cloud services run on Graviton, can we expect changes to the pricing per Capacity Unit or lower total CU consumption for the same load?

      Kind regards,

      Author's profile photo Stefan Baeuerle
      Stefan Baeuerle
      Blog Post Author

      We have indeed reduced the customer consumption rate prices of SAP HANA Cloud by over 40% since generally available. This was achieved through various technical innovations and other  optimizations.  ARM-based support by SAP HANA Cloud is one of many investments and innovations to help continue to improve overall price-performance for our customers.

      Author's profile photo William Matamoros
      William Matamoros

      Hi Stefan,

      This is available only for SAP HANA Cloud? or we can using in a TDI architecture on AWS?

      Author's profile photo Joachim Fessler
      Joachim Fessler

      Hi William,

      ARM based AWS Graviton support is only available in SAP HANA Cloud.

      Author's profile photo . Partner
      . Partner

      Hi Stefan Baeuerle

      was the HANA kernel code adapted to use AWS Graviton3 vector-instruction SIMD capabilities like it was done for Intel AVX-512 ?

      The corresponding ARM architecture is named NEON:

      There are some basic examples mentioned here:



      Author's profile photo Joachim Fessler
      Joachim Fessler

      Hi, HANA kernel code is using both NEON and SVE (Scalable Vector Extension, SIMD instruction set to accelerate some data heavy operations like scans.