Esri study shows HANA PCE in RISE supports 8x the target load in a benchmark

[TomTurchioe]

Many SAP customers are moving to RISE with SAP.  If you're reading this blog, you're probably familiar with my other blogs on SAP Community here.  If your moving your SAP ERP to RISE, your GIS system of record has to move to preserve performance and take advantage of simplified integration between S/4HANA and ArcGIS Enterprise.   Although Esri certified HANA as an enterprise geodatabase seven years ago, customers have been cautious - no utility wants to be the first one to put ArcGIS on the HANA platform.  With NiSource live in RISE with SAP with ArcGIS on HANA, now there is.

This past January, Esri ran a study that benchmarked the performance of a typical workflow mix (see the diagram below) against an ArcGIS Enterprise Utility Network geodatabase on an SAP HANA instance in the cloud.   The configuration, instance sizing, data and workflows are typical for an average mid-tier gas utility. Here are the characteristics of the dataset used in this study: a gas utility dataset with a geographic extent of roughly 55,000 square miles, and with an overall size of approximately 45 GB. The dataset contained roughly nine million features, 700 sub-networks, and 300,000 gas meters.

For some time, I worked with Esri to advocate running these studies and I'm very happy they have done them and made the results readily available here.  This blog unpacks the results of these studies and why it matters to any utility that uses ArcGIS Enterprise that wants to move to RISE. It also validates NiSource's decision to go live with ArcGIS on SAP HANA PCE last year.  For a thorough overview on this topic, read my blogs here on SAP Community on ArcGIS - SAP integration.  You can view them all here. and here.

Here are the workflows used in all the HANA studies that Esri has run so far.  While the workflows below were running, workflows were run manually to completion while the system was under load.  This was done to observe any degradation of user experience.

With RISE with SAP, there's PCE (Private Cloud Environment) which means that the SAP landscape is run in its own tenant in one of the major hyperscalars.  Esri used AWS in this and previous studies.

The study ran two scenarios:

1. An SAP HANA VM in the same VPC (Virtual Private Cloud) as the ArcGIS Enterprise related instances
2. An SAP HANA VM running in the RISE PCE tenant which is a separate VPC from the ArcGIS Enterprise instances where the two VPCs are connected with AWS PrivateLink for this study.  It should be noted that @NiSource deployed in Azure and did not need to use an equivalent dedicated link in their production environment.

This study used VPCs that are within the same availability zone (AZ) and region.  That's the only way to ensure latency is kept to a minimum.

What were the results?

The test results show that as implemented, the systems had adequate physical resources to support usage from the design load through usage that was eight times the design load…. Both systems delivered similar performance, with no significant differences in workflow times or user experience.  There was also no meaningful change in hardware utilization between the two scenarios used in the test.  Workflow response times were acceptable and consistent all the way to eight times the design load.

These results show that utilities using SAP and ArcGIS can move their geodatabases to the cloud along with RISE with SAP and know that ArcGIS Enterprise with the geodatabase on the HANA platform will perform well - in fact up to 8 times the expected load

At eight times the design load for both scenarios:

• Portal for ArcGIS and the hosting server CPU utilization generally stayed below 35% utilization
• ArcGIS Server CPU utilization generally stayed below 60% utilization
• SAP HANA CPU utilization (as reported) generally stayed below 45% utilization.  SAP provided these results obtained during the study to Esri for inclusion in this study.
• Both scenarios provide a similar user experience with negligible differences in total conducted workflow time and step completion time.

Here are the graphs showing results for 8x the design load.

Aside from the performance and scalability afforded by the HANA platform, moving the geodatabase to the HANA platform means that access to your S/4HANA Plant Maintenance master and transactional data is on-demand; the data is always current.   

If you've going to RISE with SAP and want to leverage the capabilities of ArcGIS on HANA, this blog explains why your enterprise can confidently move SAP ERP to the cloud with RISE with SAP and move your ArcGIS Enterprise GIS system of record to the #HANA platform in the cloud.

Note: the screenshots are from the Esri study linked to in this blog.  All rights belong to their respective owners.