Intel’s Response to the HP-ARM News
Radek Walcyzk, head of PR for Intel’s server division, recently called Wired to discuss the recent partnership between Hewlett-Packard and Calxeda to develop ARM-based servers. Intel would like the public to know that it will take the microserver market by storm with its Xeon and Atom.
Intel’s Ahead
One of the first things that Walcyzk told Wired was that Intel has officially been touting microservers since 2009, when the company first announced support for the idea. Back then however, the term was mostly restricted to speaking of lower-power Xeons. But this past March Intel started talking about Atom in the server space; SeaMicro had already paved the way by launching an Atom-based server product that the company had been working on since 2007. This was also after other vendors like SupeMicro, SGI, etc. had already been shipping Atom-based server products. Though late, Intel is still far ahead of ARM in the microserver market.
When you put together the huge x86-based software ecosystem with the fact that many companies have already been shipping Atom-based micro-server products for more than two years now, ARM-based microserver movement is lagging behind and will be for a while.
Also related the facts stated above is that all of the ARM microserver numbers for efficiency gains that are released are based on simulations and not actual shipping systems. Intel, on the other hand, has had two whole years to gather data from actual cloud workloads running on x86-based production systems.
Walcyzk told Wired that Intel definitely spots a huge demand for microservers. “This is a category that we think may be up to 6 to 10 percent of the entire x86 market by 2015.” And for around 66% of this growing microserver market, Intel estimates that Xeon will be the best performance per watt per dollar choice, with Atom taking over the remaining 33%.
=What’s the Best Architecture==The concept of Intel’s Xeon processors being a better fit for I/O-bound workloads will sound ridiculous to most, but the idea may not be that farfetched. Jon Stokes from Wired argues “the fact that higher levels of die-level integration (i.e. Xeon) should trump board-and chassis-level integration (i.e. a bunch of ARM chips in a rack) every time in terms of cost, efficiency, and performance.”
The verdict on what the best kind of architecture for these workloads is, hasn’t come yet. Right now, we can say with certainty that Intel’s current Xeon isn’t what the Hadoop audience wants, which is why so many got excited about ARM and Atom. But a future Xeon iteration could move into this space, provided that its designed for a significantly lower set of power and price points than the current Xeon line.
Intel might be catching out though, because earlier in March the company updated its microserver roadmap to include Intel Xeon parts that range from 45W down to 20W. Intel also plans to have a sub-10W Atom part released next year. Guess we’ll just have to wait and see what happens.
Photo by : David Bauer








| Template by