Journey of HANA from Intel Xeon Ivy Bridge to Haswell to Broadwell in an hour

Frank Stienhans

Performance of AWS EC2 Instance Types for SAP HANA

Frank Stienhans
, posted on
, posted 
October 13, 2016

What’s a Rich Text element?

The rich text element allows you to create and format headings, paragraphs, blockquotes, images, and video all in one place instead of having to add and format them individually. Just double-click and easily create content.

asdsadsadn sldjflsdjf

Static and dynamic content editing

A rich text element can be used with static or dynamic content. For static content, just drop it into any page and begin editing. For dynamic content, add a rich text field to any collection and then connect a rich text element to that field in the settings panel. Voila!

How to customize formatting for each rich text

Headings, paragraphs, blockquotes, figures, images, and figure captions can all be styled after a class is added to the rich text element using the "When inside of" nested selector system.

AWS did it again !

Just last week AWS released the EC2 Instance Type M4.16xlarge with Intel Xeon E5-v4 Broadwell.

This week it was X1.16xlarge with Intel Xeon E7-v3 Haswell. And we are still before re:invent !

The X1.16xlarge instance with 10 Gbps Network, 64 vCPUs and 1 TB RAM is in all aspects half of a X1.32xlarge with 2 TB RAM.

Something else happened: You can now get X1 instances with 1 or 2 TB RAM on the Spot Instance Market.

So I took the new Instance Type with 1 TB RAM for a spin and spent $1 for the entire test run.

In one hour with Ocean9 powered by AWS you can

  1. deploy a high end SAP HANA System, 
  2. restore a large backup with 1.2 billion rows
  3. run your SQL tests
  4. terminate the system

SAP HANA Performance across Intel Xeon generations and families

In my prior test runs I used a SQL statement which simply did not do it anymore for the X1 instances. I enhanced my reference query a bit so that SAP HANA and Intel have more to do with the 1.2 billion record set.

select avg( total_amt / trip_distance ) as dollars_per_mile from nyc.yellow_taxi where trip_distance > 0

The statement answers the question: What is the average amount per mile spent in a NYC Yellow cab? The answer is $ 5.90.

This SQL statement is all about CPU & RAM performance, as it only returns one number. This removes therefore any outbound network aspect.

Now the dataset has quite a number of records that should be filtered out. As I am testing SAP HANA Performance here, I am ignoring the data quality question.

So the filter trip_distance > 0 will ignore 8 million records with a trip distance of 0 or less (!!). But hey maybe some taxis go backward in a NYC traffic jam.

‍SAP HANA Dev / Test Configurations with AWS Cost Points

Did you see the two P2 instance types? Those were also launched 2 weeks ago. They have NVIDIA Tesla K80 Accelerators . So if you want to pair SAP HANA on Intel power with an additional 40000 NVIDIA cores, you are just one click away. Yes we are still before re:invent !

Of AWS EC2's 12 pricing models, "On Demand" is usually the least cost effective for 24/7 workloads. It can though very well be the absolute best for many workloads if you combine it with Maximum Orchestration.

AWS EC2 Instance Type, SAP HANA Performance and Intel CPU Family + Generation

The above SQL statement ran for 1.8 seconds on X1 with 2 TB RAM and Intel Xeon E7 Haswell and 10.3 seconds on Intel Xeon E5 Ivy Bridge.

For me most interesting was to see how close the new X1 instance with 1 TB RAM is to the M4 instance with 256 GB RAM. Intel Xeon E7-v3 vs. Intel Xeon E5-v4 appears to be a close call in the 64 vCPU arena.

The ECU column displays the CPU performance assessment by AWS, called "Elastic Compute Units". SAP - SAPS and AWS - ECUs usually correlate extremely well. For my SAP HANA query this is also true for Intel Xeon E5 CPUs.

X1 with Intel Xeon E7 appears to perform a bit better than anticipated by the ECU values. This is probably because the SAP HANA technology team and Intel constantly work together to get the maximum out of Intel's High End CPUs.

The Test Run

Well, I clicked six times on Launch, waited 20 minutes, used SAP HANA Studio to fire the queries and clicked six times on Terminate.

You can do that too using Ocean9.

Sign up here. Sign up is free of charge.

Ocean9 operates with Bring your Own AWS Account, SAP License and SAP Software Model.

Have a nice day.

Back to top