18 Months Of OpenSolaris Innovation

Congratulations to the entire Solaris team for today's launch of OpenSolaris 2008.11. While the credit for this amazing new release goes to many, many Sun, Intel, AMD, and other developers who worked on the release through the OpenSolaris Community, I will always consider the OpenSolaris distro one of my babies. Here is just a bit of the story.

Back in April 2007, Ian Murdock and myself, both new to Solaris marketing, took Jonathan's challenge of making Solaris more approachable and friendly to a wider range of developers and users. Solaris' key innovations, featutres like ZFS, Dtrace, and Zones were already being used back then by many Sun customers, it was just that we wanted to attract more new developers and users to Solaris, including many that were used to working with other operating systems. So off we went with a goal, to release a new OpenSolaris distribution within six months. For the first few months, all Ian and I heard, over and over again, was, "you have 0% probability of success in launching a new OpenSolaris distribution in 6 months". Luckily, a lot of Solaris developers thought otherwise.

The goals for the first release were rather simple.

  • Get it out in six months, and we set October 2007 as our deadline
  • The distribution must fit on a CD, and live boot (getting the 3+GB Solaris distribution down to 650 MB was more an internal political challenge than a technical one - oh how could you not include my favorite package in the core distro
  • The release had to be freely redistributable, we were designing for widespread adoption, not maximum sun.com registration, that would follow if we got the adoption part right
  • We would stick to a 6 month release cycle for this and future releases, release predictability and time to market for new functionality were more important than feature completeness

    It was clear from day one that Solaris' 10+ year old packaging system would have to go to support the goals. Luckily, the IPS package system was already under development by Stephen Hahn's team and provided just the trick needed. I think sometime during late summer 2007, we finally hit critical mass and had the majority of the Solaris team behind our plan. The rest, as they say, is history. Now unfortunately for a lot of Solaris engineers with school-age children, our October deadline ended on Halloween night. Around 8:30 pm on October 31 2007, Bill Neisheim typed a few final commands and the new OpenSolaris release went live on OpenSolaris.org. We all promised that night to never have another deadline on a family holiday and I still feel bad for all the Solaris engineers who missed trick-or-treating with their kids that evening.

    Before the next six month release came out, I moved back to a sales role and a real marketing professional, Jim McHugh, took over the reigns for what Ian and I had started. It was time for the startup to go public and no one was better suited to do that than Jim. The next release came out six months later, as promised, in the form of OpenSolaris 2008.5, this time launched on the new OpenSolaris.com site. With new support offerings from Sun Services, OpenSolaris 2008.5 was the first real commercial release of the OpenSolaris code base. Fast forward to today, and the launch of OpenSolaris 2008.11. Don't ask me about the 10 day delay, but I expect it has to do with Sun's busy product launch calendar these days and my earlier promise with Bill to not have engineers work another holiday (Friday after Thanksgiving was not an option!).

    Of course, a lot of developers who are not Sun employees contributed to the OpenSolaris 2008.11 release. Developers like those from Intel who ensured OpenSolaris 2008.11 was optimized for Intel's six-core Harpertown CPU, shipping today in the Sun Blade x6450 server module, a 24-core blade using four Harpertown CPUs. OpenSolaris 2008.11 is also one of the first OSs to be optimized for Intel's new Nehalem processors, shipping today in desktops and next year in servers. AMD engineers also contributed code to ensure OpenSolaris 2008.11 was optimized for their newest Shanghai processors, including those in the 4-socket Sun Blade x6440 server module announced yesterday, making optimal use of the x6440's industry leading memory footprint and I/O bandwidth. After all, who knows better than how to optimize for AMD and Intel's latest CPUs than AMD and Intel.

    I'm also really excited about today's plans announced by Toshiba to bring to market an OpenSolaris laptop. Toshiba has long been an innovator in laptops and many Toshiba innovations will be available for the first time ever on a non-Microsoft platform for developers and others who prefer to run OpenSolaris natively. Of course, Sun's xVM VirtualBox software lets developers who want to run OpenSolaris as a guest on their Windows, Mac, or Linux laptop do so, as well as supporting Windows and Linux as guests on OpenSolaris.

    I won't try to detail all the new features of OpenSolaris 2008.11 here, as the marketing team and engineering bloggers have already done a great job with that. So once again, congratulations to the entire Solaris team on today's release. I encourage everyone to give it a test drive. You won't be disappointed!

  • Comments:

    Post a Comment:
    Comments are closed for this entry.



    « June 2016