‘Power First’ As IBM Exits X86 Servers
May 19, 2014 Dan Burger
With IBM in the process of selling its X86 server business, any internal anguish caused by trashing X86 servers is removed, so the Power Systems promoters are firing away. The introduction of Power8 servers launched the volleys and, as the Power8 lineup continues to be revealed, the comparison bombing will likely intensify. Alex Gogh, vice president of server solutions within the Systems & Technology Group, talked about the “Power First” mentality in an interview with IT Jungle.
“We have committed to a Power-first mentality in the IBM software stack,” Gogh says. “This means when we release software we will release it on Power first. When I discuss the Power-first mentality, I am at the category level of Power compared to X86. Depending on the application and application layer and the application stack, the release would vary with regard to IBM i, AIX, or Linux. There are some things that will come out on IBM i prior to AIX, for example.”
Gogh’s perspective of the Power Systems market is precipitated by 30 years of IBM experience in marketing, sales, and systems development, including worldwide leadership assignments and positions with responsibilities for enterprise and SMB customers. In his view IBM i strategy is closely aligned with what IBM i shops are requesting. And it all fits into IBM’s core strategy that has been emphasized for several years.
“What’s driving IBM i today are four things: cloud, analytics, mobile, and social,” Gogh says. “We have clients that are getting hit with information explosion. We predict from 2005 to 2020 a 300X growth in data. That means, whether you are an SMB client or among the largest enterprises in the world, you have to assimilate this information at rapid rates. And the winner of that will take information as a natural resource and use if for business advantage.”
“Analytics and mobile are driving that. By 2017, there will be a trillion devices connecting to computers. That’s far greater than the number of computers that connect to one another. So the experience is going to be a mobile device or intelligent device connecting to the system. This is complemented by the ability for companies to know who I am and what I want and assimilate that information in a nanosecond.”
Of the four points Gogh and others at IBM are emphasizing–cloud, analytics, mobile, and social–it’s mobile that seems to have the best track record in the IBM i community. It’s certainly the easiest to define. Both cloud and analytics include a variety of definitions that change depending on who’s doing the talking, but Gogh sees mobile and analytics as being joined at the hip. Where one goes, the other will follow.
“What helps IBM i is the heritage of IBM i,” Gogh says. “At its core, it has always been a strong, integrated database. And now we are seeing mobile providers taking RPG code or PHP code and Java code and creating systems of engagement that connect to the systems of record. I’m particularly impressed by mobile and analytics as being major drivers.”
Bringing this discussion back to a comparison between Power and X86, Gogh relies on talking points that are traditional strengths for IBM midrange systems with a few modern transformations added in. Companies, Gogh says, want systems that are reliable, available, and secure–IBM midrange systems’ traits that have won it a loyal following. And beyond that, he adds the capability to assimilate large amounts of information, while being flexible enough to adopt open standards.
All those things are nice talking points, but they won’t win many battles pitting IBM i on Power against Windows on X86, unless you’re talking to an IBM i shop. Open standards and open source are seldom in the conversation unless there is an IBM executive or business partner doing the talking. My conversations with people in the IBM i community tend to be about what has unfolded. Gogh, on the other hand, talks more about what will unfold.
“The infrastructure of the future is going to be a hybrid infrastructure,” Gogh predicts. “I don’t care if you are talking about small, medium, or large companies. The IBM i client in the future will have on-premise capabilities that can scale out or scale up. It is truly unique in that regard. We have Smart Cloud and we have ISVs and MSPs providing their own options. There will be private cloud and public cloud environments. We think the future will be built on hybrid cloud infrastructure for IBM i and for Power customers overall.”
Price/performance comparisons are slippery fish. Whoever is doing the talking has the advantage and benchmarks are used to demonstrate superiority even though the company doing the benchmarks always comes out the winner. However, when the details of system configurations become apparent, it’s rare that apples-to-apples comparisons (our best against their best) are legit.
That said, IBM deserves kudos for its achievements in the areas of multi-threaded cores, memory management, and I/O capabilities. These are all important technologies, and they benefit Power immensely and contribute to price/performance advances.
One of the points that IBM is intent on making in price/performance comparisons is that in many cases IBM i shops will buy a single box, whereas to handle that same workload in an X86 environment will require several boxes. Sometimes that is true and other times it’s not. My advice is to beware of one-to-one comparisons regardless of who is telling the story.
I agree the opportunity for footprint consolidation is considerable. And when Gogh says, “You can take a rack of X86 servers with 200 VMs and consolidate that on one Power8 box,” I don’t disagree. It makes sense to consolidate and it makes a good IBM Power story, but the comparison against consolidating on X86 is being avoided as if it doesn’t exist. The efficiency comparisons are figured according to a best case scenario for Power with full credit given to IBM technologies such as multi-threading, memory management and the I/O bandwidths.
Those technologies are very impressive and capable of delivering tremendous efficiencies. But if the comparisons between X86 and Power and going to be discussion of price and performance, then it’s important to keep the playing field level. The best that Power can achieve compared to the best X86 can achieve. That’s what price/performance comparisons should be. And then we can decide whether “Power First” has some bite as well as some bark.