Sign up for our newsletter and get the latest HPC news and analysis.

Intel Acquires Whamcloud

Whamcloud CEO Brent Gorda writes that Intel has acquired the company.

Intel Corporation acquired Whamcloud. On behalf of Intel Corporation, I want to take this opportunity to assure you of Intel’s commitment to our current and future customers as well as a seamless transition. The Whamcloud acquisition extends Intel’s software and service portfolio in the high performance computing space in addition to reinforcing Intel’s position in the open source community. Working as one company, we are now in a stronger position to advance our mutual goals and continue providing vendor neutral solutions, delivering greater value to our customers, and moving the industry to exascale performance.

Gorda has taken a new position as General Manager of High Performance Data Division at Intel. Read the Full Story.

As for Intel, this move to bring Lustre technologies in house makes a lot of sense in their race to Exascale against Big Blue and GPFS. In the meantime, we can expect a lot of nervousness from the Lustre community, which had made moves under OpenSFS to ensure that Lustre never again became encumbered by corporate overlords as it did under Oracle.

Comments

  1. Rich

    Not sure if people are nervous about this, as Whamcloud doesn’t actually own Lustre IP. They own their own code, and Chroma. They bring Lustre development/release engineering into Intel. But Intel doesn’t own Lustre. Technically Oracle still owns Lustre.

    The argument I make is that OpenSFS was/is the central organization “guarding” the technology. Whamcloud is an instance of a supporting/engineering organization (ok, its a little stark, and whitewashes their real value). I think Intel simply wanted a seat at the Lustre table to be able to grow the value of the Infiniband assets, for areas outside of what they originally targeted.

    If Whamcloud suddenly disappeared (unlikely), would Lustre be in terminal condition? Not really, though the loss of development/hosting/engineering/support would be problematic at least in the short term. This is very different than during the Oracle acquisition of Sun.

    This acquisition also gives Intel a say in what features they might want accelerated (say multithreading the MDS, which has been a very desirable feature for a while).

    Relatively speaking, Intel acquiring a company is a good thing for that company. This is not so true with other behemoths … Anyone see Ibrix in the last few years?

    It makes sense in a bigger picture play. I tried to describe what I thought that might be here: http://scalability.org/?p=3925

    • Good point, Joe, but 98 percent of new Lustre code is written by Whamcloud and the IP could be for sale. Just saying.

  2. So Intel wants to build a Supercomputer, anyone remember the last time
    Intel was in the Supercomputing business?

    On the Lustre front… I would say this is “bad”, not as bad as the CFS/Sun/Oracle
    debacle, but “bad” none the less. 98% of the real engineering work was being
    done by WhamCloud, and while this is great for certain folks at WhamCloud,
    monetarily speaking, this doesn’t help Lustre over all in regards to the HPC
    community. Here we are, introducing uncertainty into the Lustre project, yet
    again.

    If OpenSFS was smart they would try to get more firms involved in Lustre
    engineering and support for some of the contracts/projects they have. The
    HPC community needs more then one “start up” doing the work, be that
    CFS, WhamCloud, or whoever? We need to grow a wider base of firms
    for Lustre work and projects, and I don’t mean outfits that markup solutions
    and pass the real engineering work back to the primary/single/lone company
    while skimming off the top.

    Hey OpenSFS are you listening?? Probably not, as I don’t have ~50K to sit
    at the special table. Oh well, it seems like IBM will have yet another leverage
    point against Lustre.

    • OpenSFS supports this move. This letter from Norman Morse appeared on the Lustre group over the weekend:

      “Open Scalable File Systems congratulates Whamcloud on their acquisition by Intel.

      We at OpenSFS believe this acquisition will strengthen our capability to continue serving the community. The interest of an additional major corporation in open source scalable file systems is continuing proof of the importance of these systems to the computing community. It also demonstrates that OpenSFS continues to have a key role to play in the community as it held the Lustre community focused at a time of uncertainty and will prevent uncertainty in the future. We are very proud to see that our joint work is of interest to the larger HPC community and to an organization such as Intel.

      Cray, DDN, Lawrence Livermore National Laboratory and Oak Ridge National Laboratory founded OpenSFS in 2010. The founding members have since been joined by Xyratex Corp. at the Board Level and the Board is completed by our community representative Board Member from Indiana University. OpenSFS is very proud of the many participants who have joined at levels other than the board level.

      OpenSFS is a California non-profit, mutual benefit corporation established to guarantee the continued availability of open source file systems for the high performance computing (HPC) community. Our current focus is the Lustre file system.

      Operationally, OpenSFS gathers requirements from the community for file system enhancements and support and, using membership dues, awards contracts to organizations such as Whamcloud to provide the required enhancements and support. Support for OpenSFS has grown significantly over the past year as demonstrated by increased OpenSFS membership.

      Quoting from the acquisition announcement issued by former Whamcloud management Brent Gorda, now Intel General Manager, High Performance Data Division, and Eric Barton, now Intel Lead Architect, High Performance Data Division:

      “With the support of a thriving community, Lustre is quickly becoming the preferred solution. And, with the help of OpenSFS, we are continuing our commitment of supporting all vendors with a single tree to drive the growth of the Lustre ecosystem.“

      OpenSFS will continue to work with the open source scalable file system community and with the Intel High Performance Data Division as well as other vendors to complete the enhancements and improved support the HPC community requires. We encourage all Lustre users to join OpenSFS and work with the community to ensure continued improvements.

      Norman Morse
      President/CEO, OpenSFS

  3. @Rich

    True. If they renamed it OpenCFS or OpenLFS or whatever … WhamcloudFS … IntelFS? That would be fine. I don’t care so much about the name.

    The source is open, the risk is lower, in that Intel has a pretty good presence in Open Source. I am not worried about the Whamcloud guys going away. If anything, I think the folks I know from Qlogic who joined Intel will be talking even more closely with the Whamcloud folks.

    Though let me postscript that by noting that not much has been done yet with Qlogic (at least externally facing), and this has caused some of our customers concern (those with Qlogic infrastructure).

    I am optimistic … about the acquisition. I am hoping for the best out of it.

    If it goes pear shaped (think SGE today) there are alternatives if we need, and the alternatives are pretty good, and getting better (Fraunhofer GFS) with features/functionality I am hoping to eventually see in Lustre.

  4. @tom

    I am not so pessimistic about Intel. All businesses have to make hard choices at some point, and where Intel was when they first got out of the market wasn’t a great position. All businesses go through cycles.

    Relatively, HPC has been, historically, a small market (for Intel and others). Since people are starting to get that Big Data and HPC are quite similar, I expect this view of “small market” to change, drastically. This is why I said I believe this acquisition makes sense (and several others as well). Big Data is potentially huge, and HPC’s impact upon Big Data could be considerable. All the stuff we’ve been working of for decades, in how to efficiently move data, programs, schedule computing, etc. matters when you are dealing with hundreds of TB to PB of baseline data you have to analyze. You need massive distribution, massive storage, massive networking. And the expertise to leverage this.

    I’m probably stating the obvious or preaching to the choir as it were, but I think Intel gets this too. Lots of bright people there.

    Doesn’t mean they will do HPC forever. But I think its become one of their cores going forward (though its spelled Big Data).

Resource Links: