IMHO: Sun & OpenSolaris… 2007 and 2008
Posted on December 30, 2007
Its the end of the year and I feel like some comments about where we are, meaning OpenSolaris, and where we’re headed are in order. I’ll wrap this post more tightly than I have in the past with the “IMHO” disclaimer given that my opinions are not always reflective of others. So please, interpret the following in that light.
2007 was, for Solaris and to some extent Sun, a year of transition. Looking back, the both Solaris and Sun underwent major changes 2-3 years ago, including Sun’s whole hearted commitment to X86 systems, the opening of the Solaris codebase and building a community around that code, Solaris 10 (meaning ZFS, DTrace, SMF, FMA, etc) revolutionizing UNIX, and Jonathan Schwartz taking the reigns. All those things started off strong and with force. In 2007 we’ve definitely seen things transitioning from the energy of introduction to the slowing that comes with facing the long term. Momentum has slowed and details that are sometimes overlooked in the excitement of it all start coming to the surface, often producing a “why is this suddenly an issue?” feeling.
Lets start with Sun Fire X86 servers. The initial entries from Sun way back where the V20z and V40z servers, rebrand systems that got the ball rolling. Sun’s launch of its own designed systems was an astounding no-holds-bar attack into the very competitive X86 arena with the introduction of the X2100 for maximum value, X4100 and X4200 for maximum capability, and then followed up by the revolutionary X4500 “Thumper” and the ultra scalable X4600. Wonderful systems with no compromises made. But in 2007 we got the M2 updates which broke form by trying to reduce costs in the wrong places (such as switching from all Intel Pro to splitting half NVidia Gigabit and half Intel Pro). The might have been just a simple bad decision, except that with the X4150 the trend continued. The X4150 was the triumphant result of an exciting partnership with Intel and brought the first Quad Core capable X86 systems into the Sun Fire line, but several key features of the previous systems, such as ILOM, were downgraded to the much inferior ELOM, for what can only be assumed to be continued cost savings. Whats clear is that for Sun to complete in the X86 Server space it absolutely must distinguish itself from the competition by being price competitive and constantly ahead of the game with absolute quality and technological superiority. With the simple decision to equip the X4150 with a StorageTek branded Adaptec RAID card and dumping ILOM for ELOM the divide between Sun Fire and Dell PowerEdge was slim enough that the Dell actually comes out on top… and Sun can’t afford that. Sun absolutely must push for maximum features and capability in every X86 system it builds or it will slip back off the map. Efforts to save $10 here and $15 there are likely going to cost Sun its X86 business far more than that.
On the issue of leadership at Sun, well, it would seem that more and more expectation is being put on Jonathan. I’m a staunch believer in Jonathan, a more passionate visionary you won’t find. His heart is pure, his vision strong, his banner full of hope… but there seems to be an increasing desire for a more detached business person to take the reigns. Wall streets call for increased profits is always calling and there is always pressure on a CEO, but for reasons that I can’t really explain or support, I begin to sense that forces are in motion that will mount in 2008 for change. I pray that isn’t the case. Jonathan embodies all that is great about Sun Microsystems, its not a product but a culture, and that culture is what we really love about the Sun brand, it represents something and the best and the brightest flock to it, to be part of it… Jonathan has always been ideal because while that engineering culture can’t be described it can be seen in him. Thats the nature of a brand, it’s a concept, ideal or culture thats represented by some brand and around that brand people are attracted to be a part of it in some way, large or small. If anything should happen to Jonathan I fear what will happen to what remains of that culture and the impact that it’ll have on the brand. My fear is that while someone else might be able to milk more revenue out of Sun, that the last bits of the culture we love will be extinguished in the process.
OK, that was sappy, but thats how I feel. Wipe your eyes and lets keep moving…
Looking at OpenSolaris, 2007 has been a telling year. The CAB went out, the community elected OGB came in and in the processes a constitution that wasn’t fully embraced or supported was voted in simply to get it off our backs. Lets be clear here, I don’t think many people agreed with or even read the constitution before ratifying it, rather the concern was that it was “good enough” and a failure to ratify would simply throw us back into a process not many people cared about, thus, better to vote it in, good or bad, elect a proper OGB and keep the ball rolling in the name of progress. But, the OGB has come up against a myriad of problems and setbacks, from in-fighting to uncertainty on where to go next, with a binding constitution in the backdrop.
At the same time, we’ve had some great positives, the ranks continue to grow. Ian Murdock came aboard and blew some much needed wind into the sails and thus was born Project Indiana… but as with any shake up, not everyone gets behind it. Indiana came out of the gates immature in many peoples minds and immediately met with major skepticism… but, the fact that most of that skepticism came from Sun’s own ranks was telling. Clearly Indiana represented a much larger change within the Sun organization than outside, the customer response was almost uniformly positive and met with more than a few “about time!” comments. By the end of the year most of the detractors had come around, each for their own reasons. Perhaps the largest rift that resulted was over what perhaps was the most trivial of concerns… the name. As I stated earlier the magic of a brand is the concept, idea, or culture that it represents, and here is where a trivial matter became non-trivial and emotionally charged. Questions of power plays, control, and the extent of the communities independence came to the fore front and concerns that were previously just concerns suddenly became real or perceived threats. A picture is worth a thousand words… but a brand represents a thousand ideas.
Even now this issue is unresolved. Quietly buried, for a host of reasons. Is it behind us, most certainly it is not. Is there a solution? I believe there is, and I’ve made proposals to that effect but with no success. This shadow follows us into the new year.
And the OGB elections are now swinging into motion. New board, new hope… except, perhaps not. Looking at the composition of the current board we see our best and brightest, names like Alan Coopersmith, Glynn Foster, Stephen Lau, and Keith Wesolowski… and the question has to be raised, if they can’t do it, who can? This concern haunts me, most of all. Certainly there are a variety of places in which I disagree with this board, but they are a prime group of representatives and their passion rivals my own. If Stephen or Keith or Alan or Glynn can’t wave a wand and make our problems go away, what makes me think that myself or anyone else can?
This gives rise to something I’ve proposed for some time now… the board is not enough. A distinct leader, who manages the projects variety of facets day-by-day, is required. Many of the issues that face the community are of great concern to the board but meet with questions such as “Is it our job to…” Organization of each of the community groups (CG) is an example, there needs to be hand-holding to properly organize each CG, keep tabs on them, and be that “go to guy” (or gal). The constitution lays out clear structure, for matters to be decided inside each CG, with independent memberships, endorsements, etc. Tackling this much needed work is a full time (more than that actually) effort and clearly outside the charter of the OGB, a decision making body not an enforcement group. The OGB decides things, they don’t enact them… so who does? The work is too great, I believe, for anyone within our ranks to step up and do it, not to mention concerns some people would have regarding that persons credentials or authority to do so. A leader is needed to pick up where Jim Gris left off and drive this community forward. To put this in other terms, we have a congress, and they can act as a sort of court, but we lack a president. Decisions can be made all day, but there is no one to carry them out or enforce them, and the project is just too vastly huge to assume everyone will jump on board without a decisive macro leadership.
And so 2008, in many respects, enters with a sigh rather than a rush of excitement. Will Sun, will OpenSolaris, will Solaris upshift into the next gear or continue to rev into redline? It is clear to me that this last year we’ve allowed things to settle a bit, what few pushes happened haven’t panned out, and we’ve got to pick ourselves up, dust off the complacency, bitterness, and frustration and step back, re-examine the problem with fresh perspective and new enthusiasm and get back to pushing and innovating and striving for something better.
In 2008 we set a course for the future… we seal our fate or we cherish and propagate our fine culture. Engineers are being torn away from Sun and it remains divided against itself. Let not that statement be repeated 12 months from now!