Power Systems from a Competitor’s View

Edit: Why wouldn’t you run POWER?

Originally posted July 5, 2016 on AIXchange

I’m always interested in stories about customers that choose to migrate from x86 to POWER8 systems. When I hear about 2X performance compared to x86 when running workloads on POWER, I wonder how anyone could consider anything else. Throw in the AIX, IBM i, and Linux on Power operating systems, and to me there’s utterly no reason to run another operating system on other hardware.

Of course, IBM’s competition will make their own cases. Via Twitter, I found this document on migrating from Power Systems to HPE Open Systems, and I’m legitimately curious to hear your own responses to these arguments:

“Hewlett Packard has several decades of experience in migrating mission-critical applications from IBM Power Systems to HP (and now HP Enterprise) open systems. HPE has demonstrated that the majority of such migrations result in a significantly less expensive operating environment – often by a factor exceeding 50 percent. At the same time, the new HPE open environments match or exceed the performance and availability attributes of the original Power Systems.”

First they talk about fewer ISVs supporting POWER. Then they discuss costs.

“Of special importance is the cost of the Oracle database management system. Many of the applications being migrated use an Oracle database. Oracle charges twice as much per CPU for Power Systems than it does for x86 platforms. Furthermore, Oracle RAC (Real Application Cluster) costs $11,500 on an x86 and $23,000 per core on an IBM Power System.”

Why does Oracle charge twice as much per CPU? Because POWER can do twice as much work, so you need half as many cores to run your workload.

It might be worth reading through these slides and thinking about why infrastructure matters and why you might consider POWER systems. Some of these same concepts were covered in a recent AIX Virtual User Group session (video here).

POWER8 has 4X threads per core, 4X the memory bandwidth, 6X cache per core, and runs at higher clock frequencies. Performance per core has grown with each POWER generation, which means you need fewer cores to do the same amount of work, and you can consolidate more workload onto the same server.

Why is Google interested in POWER servers? Why are these new high performance computing contracts being won by POWER servers?

“The other reason to think that Google is serious about the OpenPower effort is that Google is a big believer in the brawny core – as opposed to the wimpy one – and the POWER8 chip has more threads, more memory bandwidth, and will have several high speed interfaces, including IBM’s CAPI and Nvidia’s NVLink, to attach adjunct memory and processing devices directly into the Power chip memory subsystem and processing complex.

“There are only two brawny cores out there: Xeon and Power,” MacKean explained. “Power is directly suitable when you are looking at warehouse-scale or hyperscale computing from that perspective. I actually think that the industry is going to be moving towards more purpose-built computing, and I think that is different users are going to be able to leverage the advanced I/O that IBM is opening up through OpenPower. They are going to be able to go with purpose-built platforms that suit their workloads. I think this is a big part of this. We just heard about the CAPI 2.0 interface having twice the bandwidth and we are actually excited about how that will play out at the system level. It is open, and we are seeing a lot of people innovating in a lot of directions.”

Google gets it. When you’re running critical workloads, you’re not looking for ways to cut corners.

Where do you see yourself going in the future?