You are currently on IBM Systems Media’s archival website. Click here to view our new website.

POWER > Systems Management > Data Management

Simplify and Accelerate SAP HANA Landscapes

SAP HANA


Since the initial release of SAP HANA for Business Warehouse on IBM Power Systems* in 2015, client adoption and the breadth of supported technologies have been growing rapidly. This article covers the most important variants and experiences of the current SAP HANA on POWER* offering.

Most POWER8* and POWER9* processor-based systems using the IBM PowerVM* hypervisor can host up to 16 production SAP HANA databases (differs per model) along with their surrounding ecosystem, like application servers, preproduction SAP HANA instances and third-party software (SAP Tech Note #2230704). Architecting an end-to-end SAP HANA landscape onto a fully virtualized Power Systems platform allows agile resource adjustment among the co-hosted applications.

These consolidation scenarios help in fully exploiting the huge physical memory, processor capacity and virtualization flexibility in POWER9 systems. Clients who prefer a scale-out topology, or where data footprint mandates it, can deploy up to 16 POWER-nodes as a scale-out cluster, or even better, collocate multiple scale-out nodes as LPARs on high-end Power Systems to reduce overall physical footprint and complexity.

Scale up for SAP HANA

That said, the clear recommendation is to use scale-up deployments at a maximum of 16 TB for Online Analytical Processing (OLAP) like SAP BW/4HANA and 24TB for Online Transactional Processing (OLTP) whenever possible (SAP Tech Note #2188482).

Clients get excellent performance—and far less complexity—with all data contained in a single memory space, instead of being distributed across multiple database nodes, which must be kept synchronized and accessed through a dedicated intra-node network. Massive parallelization of query processing on CPU level, which is generally considered an advantage of scale-out implementations, can be maintained by logical partitioning large HANA tables within scale-up LPARs.

SAP HANA provides intelligent policies for data partitioning supporting high concurrency for queries scanning large numbers of data records. In many cases, clients migrating to HANA on POWER scale-up systems from earlier HANA appliances clusters see significantly faster processing of their SAP application workloads.

The introduction of flexible Tailored Datacenter infrastructures (SAP HANA TDI) and workload-driven sizing methodologies with TDI Phase 5 allow clients to fully exploit the versatility of the IBM POWER platform in various aspects:

  • Scale up to a maximum of 192 cores/ 24 TB per single database LPAR/node
  • Scale out to 16 physical or virtual nodes for OLAP, up to 4 nodes for OLTP
  • Customized SAP HANA workload-based sizing, leveraging SAP provided reports that analyze existing SAP applications and data structures
  • Consolidation of up to 16 production SAP HANA databases on a single E980 server
  • Use shared processor pool for SAP HANA database workloads for effective and efficient utilization of CPU resources
  • Attachment of any vendor certified storage subsystems for HANA data persistency

OS Considerations

At the OS level, clients can choose between SUSE enterprise Linux* 12 and above or Red Hat Enterprise Linux 7.4 and above. Both distributions require specific versions for SAP HANA, which are tailored to exploit IBM POWER and SAP HANA technologies with best reliability and performance. In addition, they provide the technical foundation and support service-level agreements required for running mission-critical enterprise applications using an SAP HANA database.

They complement SAP HANA System Replication (HSR), which is an inherent high availability and disaster recovery solution used with scale-up solutions. HSR takes care of data consistency between primary (active) and secondary (standby) HANA systems, basically through the means of transaction log shipping. SAP HANA on POWER supports all HSR modes and thus any recovery time objective requirements clients have. The Linux HA extensions manage the automated cluster takeover on infrastructure level. Optionally, IBM PowerHA* high availability or third-party solutions can be used as cluster software.

More than 2,100 clients today are taking advantage of POWER to simplify and accelerate their SAP HANA environments. Many clients moved from x86 to Power Systems servers, recognizing the advantage of unmatched scalability along with the flexibility offered by PowerVM virtualization. It enables them to provision SAP HANA instances faster, size based on the actual business needs, leveraging its ability to do granular resource allocation and operate with high reliability.



Advertisement

Advertisement

2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

POWER > SYSTEMS MANAGEMENT > DATA MANAGEMENT

Are You Ready for GDPR?

POWER > SYSTEMS MANAGEMENT > DATA MANAGEMENT

IBM Researchers Maximize Apache Spark’s Capabilities

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store