Is PowerLinux Relevant? The Case for PowerLinux

March 12th, 2014
Ron Gordon
Director of Power Systems

“Is PowerLinux relevant in the Open Source and Linux solutions market place?”

That question is asked by customers planning application implementations and also many analysts and ISVs. In my opinion, the answer is: absolutely yes!

I could leave it at that, but you may want to understand my reasoning. If we back up and take a look at one of the original premises of Linux we start to see where OpenPower Linux fits. Linux, in my opinion, was developed (besides a cool university project) as an operating system independent of underlying hardware technologies. As such, if you compiled the Linux build for x86, it ran there; if you compiled it for Power, it ran there, etc… Also there were 21 “architectures” originally supported to provide an OS that basically ran on whatever you had, either client systems or servers that could actually support underlying hardware features that might be unique to the architecture.

It seems that many of the architectures have gone away (MIPS, ALPHA, etc) and there are really only three meaningful chip architectures today…x86 (and all its versions, like 32 bit, IA, 64 bit, etc), 360 Principals of Operation (aka System z) and POWER. As Linux really runs on all of these (through various distributions like Redhat, S– USE, Debian, Ubuntu, CentOS, etc.) with the same levels of APIs and systems functions universal, I think the value of PowerLinux is not in Linux, but rather in the hardware architecture and what applications really run best and most economically on it.

Power fits very nicely between x86 and System z. It has more performance, capacity and reliability features than x86 servers and is just under System z in reliability and IO performance. So, if you want to run a single purpose application that needs low performance and minimal reliability, I really would suggest an x86 system for cost reasons…applications like firewalls, sprayers, LDAPs, etc are in this category. Also, if you are a small business, for example (as the old commercial stated, Mom’s Rugs…she only had one rug to sell) even web commerce and a web presence is fine on x86.

But let’s move up to where the system truly is the company and its life, such as a regional bank that needs a demand deposit system, or a manufacturing company of 10,000 auto parts that needs an SAP system for life cycle management with reliability and performance. In this case, OpenPower would provide a better solution, probably on a single enterprise class Power System as opposed to 10 or more x86 servers and most likely at a lower price too! Now if a very, very large bank wants to run a worldwide program for customer wealth management, the magnitude of the solution and reliability requirements may be better suited to System z. All are Linux, but the solution requirements dictated the best hardware support and the corresponding TCO.

Now you might say that cost usually outweighs many factors, but IBM has taken tremendous steps to put the price point of OpenPower right at that of x86 systems, and in many cases potential customers are not aware of this.

Mainline