Sunday, March 9, 2008

From higher altitudes!

I took a weekend break in Boulder. The first for many weeks, and I needed it. I had just returned from Orlando and on Sunday, it was back on the plane for a trip to Prague – one of my favorite cities in all of Europe.

The weekend along the Rockies proved to be unseasonably warm so it was a good time to grab the motorcycle for a ride up into Wyoming and to revisit some of my favorite front-range roads. The picture I have included here is of me parked at a rest stop just outside Laramie, as Interstate 80 (I80) crests the 8,600 foot summit – the highest pass on I80 as it connects New York with San Francisco.

While the temperature for most of my trip had been in the 70sF (20C), passing over this summit brought it down to the low 40s (5C) with winds pretty constant at 45 mph (60+ kpm). It was also at the midpoint of the 300 mile (480k) loop I had chosen and the views from this altitude were spectacular. I took the time to have a good look at the surrounding mountains and to peer down into the alpine pastures and valleys.

At one point on this ride across Wyoming I had an anxious moment. I just didn’t know for sure where the next town was and, with no gas gage on the motorcycle, I was not quite sure if I had enough gas to make it into Laramie. As luck would have it, I came across the town of Buford and it had a gas station. It has a population of 1, making it the smallest township in the U.S. Finding it was completely by chance and I am sure I would have been less stressed if I had have consulted a map before leaving Cheyenne.

I suppose it’s not the smartest thing to do, but I never take maps. Big-twin cruiser riders (as opposed to our touring brethren) never need maps – it just doesn’t look good to be seen stopped along the side of a highway, checking a map, and looking lost. We just keep riding till something looks familiar, and we don’t turn around just to be safe. And we never, ever, stop at gas stations to ask directions!

As I was sitting in a Starbucks in Ft Collins I heard a familiar song, by Jim Morris, called Southward where he sings “It was chilly through Virginia, but I hit the Outer Banks, and my attitude was toasty as a fire … Took the Cedar Island ferry on a morning bright and clean, what’s behind me never crossed my mind at all.” Listening to the music also reminded me of something similar from the ‘70s movie - Gumball Rally. In perhaps the most memorable scene Raul Julia, playing the Italian race car driver, Franco, delivered his famous one-liner “the first rule of Italian driving?”, as he tore off the rear-view mirror, “what’s behind me is not important!”

Yes, I had ridden across mountain summits and I had been cold. I had taken time to enjoy the scenery from that altitude. But now, with my latte and a warm corner to curl up in, what was behind me was no longer important and riding without a map hadn’t proved disastrous. As I mellowed, and began to anticipate the next leg of my ride, I began to think about all the roadmap presentation I have seen of late.

I had sat through roadmap presentations on NonStop at SATUG, as I had sat through Intel’s chip roadmap. I had sat through IBM’s plans for the System z as the z10EC mainframe rolled out last week. I had sat through many GoldenGate presentations as well. Yes, product roadmaps are very important, and they provide a wealth of information if you just know what to look for! There’s not a product manager on earth that doesn’t get passionate about something they are directly involved in, and they are always eager to press home the point of how such-and-such feature is going to dramatically improve the fortunes of the user, the technology, and their own development team.

And it reminded me that sometimes, watching presentations on product roadmaps, there was much to be learned from what was not said as from what was actually said. Likewise, you could also learn a lot from recalling what had been included in previous presentations, but had never materialized as product. In today’s litigious world, what actually makes it onto a roadmap has been so well-sanitized that it often lacks drama or excitement. Resources are committed - there are real dates, test plans are in place - and you can anticipate an early adoption program following shortly.

So much sanitizing that you just have to wonder what else is in the product and technology pipelines. For me, there’s always the nagging suspicion that there’s some really incredible technology being worked on that could just tip the scales a lot further in favor of whatever product is being presented. While I am beginning to develop my own “wish list” roadmap for NonStop, and pepper it with my own outlandish requirements, I keep tripping over the non-appearance of items I really thought were on the roadmap.

From the time Martin Fink stepped to the podium in Berlin to address the European ITUG event, it was clear that the paths of the Open Systems / Linux and NonStop would intersect. At first, I took his comments to imply that at some point, we may have elements of NonStop integrated with Linux. Something modeled on the work first done during Bill Heil’s watch - the ServerWare (later NonStop Software) project. Back in the mid ‘90s a lot of energy went into overlaying Windows with enough of NonStop’s message system to provide Microsoft with the foundation for a clustered fault tolerant offering. Perhaps, what went on before should be considered more seriously. Perhaps what was behind us remains relevant, after all!

While this lab project never saw the light of day, and ServerWare development has been abandoned following the Compaq acquisition, I suspect that much of what was learnt has now been revisited but this time with a focus on Linux rather than Windows. While Martin may not bring a fault-tolerant Linux to market I wouldn’t be at all surprised to see some lower-level components and routines within NonStop draw from Linux, and the experiences gained from ServerWare.

Another similar project I really thought was fully funded was what Cupertino internally referred to as Hybrid Super Clusters (HSC). Fueled by what some of the biggest customers had been doing (e.g. Sabre, as well as some stock exchanges), and where a mix of NonStop, Linux and HP-UX were deployed essentially as clusters, there was some consideration given to productizing some of this. As Martin presented his vision of “shared infrastructure blades” that supported a mix of blades and operating systems, I thought that this was really a sign of what was to come. I could envisage a marketplace that really could take advantage of such a package as and when it materialized.

Turns out, this was never on the product roadmap, in any official sense. From the lofty heights HP NonStop executives move in, this was more a framework for discussion than anything else. Or, was it? Recently, I was taking to a development executive who pointed out to me that “work (internally on HSC) has now been completed … one of the values that resulted from this project was our recently released NonStop Cluster Essentials product; it provides an integrated management interface for monitoring, managing, and controlling heterogeneous clusters of NonStop and Linux.”

This development I find very important. In the roadmap presentations given at SATUG, a senior product manager broke down the key areas of development into just three categories – Open Access, Scale and Availability, and Manageability and Compliance – and where the issue with Manageability and Compliance was the need to support cross platform deployments. We know that specialty controllers will begin to appear shortly – Linux on blades hosting communications and other I/O controller software - and for them to be successfully deployed, they would have to be seamlessly integrated into the overall manageability scheme.

Perhaps this is where we begin to see the pieces all fitting. As we see the Intel roadmap being followed, with dual-core, quad-core, and eventually multi-core chip technology being supported, we could see the morphing of NonStop with Linux appearing in a more subtle ways – integrated as part of the lower-level components of NonStop, as the OS of specialty controllers, and all managed from a single, integrated management environment. Is this what we expected after hearing from Martin? Is this the kind of Hybrid and Hybrid-Clusters we were anticipating? Again, not really! But on closer examination – is this of more immediate value. Absolutely!

As I looked down from the mountain pass I couldn’t always see the road behind me. And I couldn’t always see what was happening in the valleys below. I may be reluctant to pull out the roadmaps while riding a motorcycle – but when it comes to technology they are incredibly important. We may not have visibility into every project the labs are working on and we may not see everything appear as a supported product. But where these roadmaps direct us, and the history that comes with them, gives us a pretty good idea of what to expect.

Projects, whether on the roadmap or not, always are leveraged at some point and the products we finally see being delivered frequently employ components and routines reminiscent of earlier lab activities. I can only now speculate that when we see future bladed architecture product offerings then, to paraphrase the Rolling Stones, “we may not get what we want … but we may just get what we need!”

No comments: