IBM Power Systems Review

Virtualizing, we're able to share across multiple organizations, push utilization over 70%

What is most valuable?

  • HMC
  • PowerVM

How has it helped my organization?

It allows us to share the resources with multiple lines of businesses. That's one issue we had with one line of business, purchasing a physical server then it would be dedicated to that line of business. There would be one OS on it so a lot of the resources were not utilized. Now with PowerVM, we're actually able to sell them the LPAR itself and the corporate entity can purchase the physical asset. This allows us to push the utilization up to 70 to 80%.

What needs improvement?

I think IBM needs a little more work on managing the overall environment with eliminating Systems Director. They need something that you can use to manage the entire environment; it's kind of where they're going with PowerVC, but with the POWER5, 6 and 7 they're out in the cold now. It's just upgrading to 8 and managing everything with PowerVC, then it will be a lot easier. But any of the older technology is going to be out in the cold, managing one at a time.

For how long have I used the solution?

We've been using Power for 10 years. We're running versions 5 through 8.

What do I think about the stability of the solution?

Lots of issues, but with hardware an issue we faced with our POWER7s has been the voltage regulators. IBM has been pretty good about that. They've been keeping voltage regulators onsite, so that if we do have an issue they're able to replace them in an expedient amount of time.

What do I think about the scalability of the solution?

It's very scalable. That's one of the advantages of Power, the ability to isolate every LPAR, whereas with Oracle using the containers, you have a global container, so it's difficult to segregate those. The way the Hypervisor does it on Power, you can actually have PCI and non-PCI on the same physical asset and still maintain PCI compliancy, but on x86, on Oracle, you cannot do that.

How is customer service and technical support?

Our SSRs in our primary datacenter are fabulous. We have one SSR, he's onsite almost every day. We actually have an office for him whether he's working on our work or he's working on the other customers that he supports. It doesn't matter, we give him the access to our site too. He's a very valuable member of our team, even though he's an IBM SSR.

Which solutions did we use previously?

We were a large Solaris shop, so we had a lot of Sun and we outsourced to IGS. When IGS came in we started replacing a lot of the Sun with Power.

How was the initial setup?

Hardware upgrades, now that we're with our virtualization, is pretty simple. We just LPM from one to the other. 

The software is a little more complex than I think it should be. I think there is some stuff that they could do with the patch bundles. They call it a patch bundle, but really it's not a bundle. There are a bunch of patches there, and you have to do an MGET and get all of them at once instead of it being one tar bundle, and you just download that tar bundle and then untar it. Then you have them. If your LTP fails during the download, it's like, "Well I didn't get them all, so, which one did I get?" Let me just erase everything and restart. 

I'd rather just grab a tarball and untar it and that way I'd have the readme right there in that uncompressed location. 

It's some of the stuff that they have like their VIO, I just downloaded the VIO DVD one, DVD two; I think it's the expanded tool kit. They're all compressed differently. One is a raw ISO, one is a compressed ISO, and the other one was a gzip tar file. I'm thinking, "Why aren't they all the same?"

Some of that gets a little irritating but you just have to deal with it and, hopefully, somebody will realize it and fix it.

What was our ROI?

In terms of the upgrades, moving from previous versions to POWER8, I absolutely see a return on investment. We're virtualizing it, and being able to share across the multiple organizations that we support, we're able to push the utilization upwards of 70%. 

Previously we would create physical LPARs and there would be one or two LPARs and we'd only be utilizing 10% of a 770, or the 570s, or 670s. So it was a got a million dollar system, and we were using 10%. That's $100,000 worth of use, $900,000 is not being used. 

Now we're pushing that utilization to where we have a lot more virtual LPARs and we're actually using that full system instead of having ten million-dollar systems. We have one million-dollar system and we're using ten virtuals on it.

Which other solutions did I evaluate?

We use competitors, Intel-based Linux.

What other advice do I have?

IBM is moving faster with their improvements than we can implement. Coming here, to the Power User Conference, to learn the new features means I then go home and try to implement this feature and see how we can actually make this a value add for our organization.

Power is the best. There's not much that can beat the way they virtualize it. And the HMCs, being able to manage the entire environment.

They're definitely a leader. They lack the advertisement to new corporate CEOs. You're starting to see more advertisements of Watson. But AIX... the Power environment and the value add that it has over Intel, not so much. Everybody thinks that Intel is so much more cheaper than the IBM, but it's because it's not marketed correctly. 

With Power servers, you get so much included with your purchase. You get the virtualization, you get the operating systems. Whereas, with Intel, you get hardware and then you have to add all of the operating systems, the virtualization if you're using VMware. And once you start adding that up, that commodity server is now only hundreds of dollars difference from an IBM server. A lot of corporations aren't looking at it that way.

Disclosure: I am a real user, and this review is based on my own experience and opinions.
Add a Comment
Sign Up with Email