SAS is a leader in the data analytics world. Its mission is to provide useful and timely insights enabling businesses to make the right decisions. HPE has partnered with SAS for decades and together, we offer world-class solutions for our joint customers. This blog highlights how SAS 9.4 gets powered by the extreme performance and resiliency of HPE Alletra storage offering full control on your data with cloud-based management and as-a-service consumption.
Any enterprise-grade analytics environment involves hundreds of data analysts running thousands of jobs at the same time.
HPE Storage has been at the forefront in serving such performance-heavy workloads. HPE Alletra 9000 and HPE Alletra 6000 storage lead the HPE Storage portfolio offering extreme performance and ultra-low latencies while running SAS 9.4. On an average, SAS 9.4 requires a minimum of 150 MB/s per core of throughput. Hence, based on the total number of cores needed in the compute layer, the required amount of storage can be calculated.
SAS 9.4 on HPE Alletra 9000
HPE Alletra 9000 is an all-NVMe, AI driven data infrastructure that offers extreme performance and 100% guarateed data availability. Figure 1 shows SAS 9.4 performance on a 4 node HPE Alletra 9080.
SAS is a leader in the data analytics world. Its mission is to provide useful and timely insights enabling businesses make the right decisions. HPE has partnered with SAS for decades and together, we offer world-class solutions for our joint customers. This blog highlights how SAS 9.4 gets powered by the extreme performance and resiliency of HPE Alletra storage offering full control on your data with cloud-based management and as-a-service consumption.
This graph shows the SAS 9.4 read+write throughput values against the number of concurrent users or jobs. This benchmarking is done using the SAS Mixed Analytics workload to simulate a real world analytics environment. Here the total number of users are divided into sets of 30 and each set runs 102 jobs. These jobs vary in duration from 50 seconds all the way up to 20 minutes. Each of these durations are depicted with a different color on the graph. A line crossing the graph depicts the total number of concurrent analytics jobs being run by the users in each test.
How to read this graph
To read this graph, the first thing you need to figure out is how many concurrent users or jobs are going to run in your analytics environment. Once you have that number, then you need to know the average duration of these jobs. With these two values, you can select which part of the graph you should look at to determine the throughput that you can expect from SAS 9.4 running on this infrastructure. In this setup with SAS 9.4 running on four HPE ProLiant DL580 servers, HPE Alletra 9080 offers a maximum read+write throughput of 29 GBps. More information on this benchmarking is available in the Maximize SAS Analytics 9.4 performance with HPE Alletra 9000 technical paper.
SAS 9.4 on HPE Alletra 6000
HPE Alletra 6000 is also an all-NVMe storage array that offering accelerated performance and 99.9999% data availability guarantee. Similar to HPE Alletra 9000, we have benchmarked HPE Alletra 6000 storage arrays with SAS 9.4. Figure 2 shows how an HPE Alletra 6070 performs with SAS 9.4:
As described above, this graph also depicts the throughput obtained for a given number and duration of concurrent jobs. With three HPE ProLiant DL560 servers, HPE Alletra 6070 gives a maximum read+write throughput of 13.5 GBps with SAS 9.4. For details, please refer to the Accelerate SAS Analytics 9.4 performance with HPE Alletra 6000 technical paper.
Sizing SAS 9.4 infrastructure
These graphs can be used in two ways, to arrive at the right performance expectations, as described above and to size the compute infrastructure. Since SAS 9.4 performance is a factor of the total number of cores in the compute, once you figure out the throughput based on the number of concurrent jobs, you can use the 150 MB/s per core requirement, to arrive at a number of compute cores needed. HPE Compute offers a plethora of server models like the HPE Superdome Flex 280, HPE Synergy, HPE ProLiant DL380 Gen 10, HPE ProLiant DL580 Gen 10 and more. You can select any of these compute options once you know how many cores you need.
As shown in Figure 3, HPE Storage portfolio for SAS 9.4 HPE Storage can serve any SAS 9.4 performance requirement from the lower end of 3-5 GBps all the way up to 30 GBps.
The HPE Storage advantage
Apart from performance, there are several other benefits that HPE Storage offers. HPE GreenLake for Block Storage is the industry's first block storage as a service offering with self-service agility and 100% data availability guarantee for mission-critical workloads like SAS. HPE Alletra 9000 and 6000 can be consumed on-premises within your data centers, as a service.
Another advantage you get is cloud based management. Data Services Cloud Console offers unified data management capablities using a SaaS-based console, bringing the cloud operational model to HPE Alletra arrays.
This video offers more insights into HPE Storage solutions for SAS 9.4.
HPE provides holistic, end-to-end solutions for SAS Analytics making sizing, deployment, and management of infrastructure highly straight-forward and easy. Please consult with your HPE representative for more info on HPE Alletra storage.
Anshul Nagori
Hewlett Packard Enterprise
twitter.com/HPE_Storage
www.linkedin.com/in/anshulnagori/
hpe.com/storage
Anshul_Nagori
Meet HPE Blogger Anshul Nagori, Senior Worldwide Technical Marketing Engineer. Anshul works for the worldwide storage solutions team at HPE, and has more than 13 years of IT experience. He is a regular speaker at SAP and HPE events. His areas of focus include SAP HANA, SAS Analytics, storage, data management, and data protection solutions. Connect with Anshul on LinkedIn!