Thursday, December 18, 2008

Still the need for something special!

I drove this past weekend from the West Coast back to my Boulder, Colorado home. For me, driving is not a chore and on a recent company conference call, I talked of my preference for driving, much to the amusement of my colleagues. Time permitting, if I need to attend a meeting and it’s less than 1,500 miles away, I would rather drive. Even with yesterday’s gas prices, staying away from airline travel had its upside and I am happy to report that it cost me a third of what it did a month ago to fill the SUVs tank. It was still nearly twice as expensive as if I had taken my daily-drive for this ride!

However, I can’t always count on the weather – and this time, it hit with a vengeance and the picture above is of the SUV the morning after I arrived home. I started the homeward leg early Saturday morning, in San Jose, with the hope that it would be an uneventful trip and that the weather would hold off for a few days. I also hoped that taking time out to visit friends just past Sacramento wouldn’t hurt.And then the adventure really began!

As I drove west on Interstate 80 (I80), there were signs warning of deteriorating road conditions and to be prepared for winter driving in the Sierras. Chains would be required except on All-Wheel-Drive (AWD) vehicles fitted with snow tires on all four wheels. I dialed in the local radio station for further updates, and it looked like it was going to get nasty. So I made sure I left my friends place before lunch – I had planned on driving all the way to Elko, Nevada.

Readers of this blog will know that my favorite cars are those with two seats, a big V8 up front, and power delivered through the rear wheels only! But for this adventure, I was glad I had an SUV with good winter tires on all four wheels. The drive over the Sierra’s presented no immediate problems but the temperature plummeted and by the time I checked into the hotel in Elko, air temperatures were already hovering around zero Fahrenheit.

But Saturday was just an introduction and did little to prepare me for Sunday’s drive!

Only an hour or so out of Elko, on Sunday morning, and warning signs were flashing alongside I80. Tuning into a weather channel, I found the road was closed the other side of Salt Lake City, just as it crossed into Wyoming. Temperatures were now -10 Fahrenheit. I have never “bonded” with the SUV – more or less putting up with it as I lived in Colorado - but I was beginning to welcome its capabilities as I turned off I80 and headed south on I15.

By late Sunday it was looking bleak. The drive home was deteriorating rapidly as now, having turned East onto I70, and heading into the heart of the Rockies, I encountered snow covered highways with temperatures steadily dropping. At one point, I saw -35 Fahrenheit as I crested one of the summits. The roads were slick and traffic had slowed to a crawl. Big-rigs were off the road, and those attempting to make it through the passes were heavily chained. It was a time for specialized AWD vehicles and no place for the regular daily-drive family car.

Finally, after 14 hours behind the wheel and with 850 miles covered that day, I turned into my driveway.

Rising early Monday morning, I was catching up on email and checking the latest postings on a number of community sites, when I came across an exchange that reminded me of the weekend’s drive home. Posted on the Connect Community web site was the question “NonStop: Do you think, that new object oriented, modern systems can run on HP-NonStop by using java?” that generated a number of exchanges between a couple of very active members from the NonStop Java SIG.

And buried within the exchanges were three observations that brought driving the SUV into focus.

The first item was “often new projects start on other hardware systems and not on HP-NonStop. And this is the problem of HP-NonStop. We have to tell them, that the new HP-NonStop is modern, too. With the new NonStop hardware, you can start to modernize your old systems. This means, not only the client dialogs. The server side can also be object oriented.”

A little later I read “if I see the German TV spots of IBM, they tell all people how good and modern they are in a funny way. Only very rare and seldom you see, if you have luck, a HP-NonStop advertising in some special IT magazines. We need more and better marketing to get the interest of young people and CEOs to get new customers and java projects.”

Finally, I saw this short piece “what do you think about the HP-NonStop marketing concept? How can we reach the young people? When I finished my diploma at university, I never heard from Tandem (now NonStop). Until now, I believe that this situation doesn't change. Our German students never heard from NonStop, but they know many of the other unix systems and works with them at university. There comes a time, when these young people will be decision maker and they will decide to use their known hardware.”

Evangelizing? Advertizing? Educating?

We need to be careful when evangelizing NonStop, as it’s not always the most appropriate option. While SUVs are viewed as excessive and wasteful, there are situations where they prove to be valuable, and they can work well in getting you home safely. But as regular daily-drives? Not a smart choice! And isn’t this the same with NonStop? Should we even be considering NonStop as a regular general-purpose computer? Isn’t it better suited for solutions that need the NonStop advantages?

And its “popularity” has nothing to do with HP’s marketing efforts or lack of educational support at our universities. Consider IBM’s System z where there continues to be a lot of advertising supporting the platform, and a major push into universities, but when it comes to deploying “modern applications” it’s rarely considered, other than in combination with zLinux. It’s not at all clear to me whether the efforts exerted to date have produced any results at all!

There’s no dissention on my part when it comes to evangelizing NonStop. Just as I have no issues with including NonStop in advertising and education programs. There should be no elements within the broader IT community unaware of how “modern” the NonStop platform has become, and how easy it is to deploy modern solutions on NonStop. But where in today’s IT architecture does NonStop truly fit – and shouldn’t we only advocate when we do see a fit?

This is a really important issue for me, as it cuts to the very core of why NonStop! I contend that it’s premier role remains deeply rooted in transaction processing, and in all my years in IT, I have never encountered a platform better suited to transaction processing. The ease with which online transaction processing solutions can be implemented is still unmatched by any other platform – and this should not be ignored as we architect solutions for the next decade. And I am pleased to see Pathway making a comeback in the 21st century – even if just to underpin other key subsystems like Java containers and SOAP. With so many discussions being centered on cloud computing, and treating all the resources within the cloud transparently, surely the NonStop platform’s attributes position it as an ideal front-end. Once again!

Let me not be misunderstood here – and I have been accused of holding my punches on some occasions – transaction volumes will only continue to go up, and with increasing reliance on the web as the connection fabric there really can be no limits set for how many transactions any application will face. Nothing is as available as NonStop. Nothing scales like NonStop. And there’s no place more suited for this than front-ending today’s business logic and data.

NonStop is not a general purpose computer and never will be – the current hardware packaging still comes with a premium. Will “blades” change the economics? Potentially – but only if the major operating systems targeting blades agree on a single piece of real estate, including the choice of interconnect technology. ServerNet needs to be replaced with something more universally accepted by the industry. The price will only reach parity when the basic blade hardware is completely standard for all of NonStop, HP–UX, and Linux. However, I do see this all happening in the not too-distant future.

As I drove over the snow-swept Rockies and battled through the poor visibility I encountered, it was good to be in a car built for the conditions. However, I never did loose sight of its drawbacks and was happy to know I had another car as my daily-drive. And this is the message of NonStop – no matter how evangelic we become, or how big our education and marketing budgets grow, we need to pick our fights carefully and approach new opportunities for NonStop with caution.

After all, there are many circumstances where ignoring the general purpose, and going for something more specialized that you can count on, certainly brings with it piece of mind!


Anonymous said...

Richard, you can add some up-and-coming 'open' environments (Spring and Hibernate) to the ones you mention that make it even easier to target NonStop - since the location where objects run is moved completely to the Administratration group and away from the developers.

I could see this as a way for the IT management to choose the correct platform (and even change it based on lessons learned!).

Bill Honaker
XID, Ltd.

Richard Buckle said...

It's probably no secret by now that in my last days at ACI I took a look at the different "supported stacks" in support of open source - with the idea that perhaps ACI should be leveraging more from them. (A casual survey found many projects within ACI already using some open source with folks already asigned to supporting it. Lot's of folks!)

But then along came IBM - and the discussions, while I was there, turned from open source stacks to blue stacks. Oh well ...

The point is however, and just as you observe, if the port is robust and inherits enough of NonStop and transparently, then this could really stir things up and have folks re-thinking NonStop for sure!


David Finnie said...


Agreed completely - and I like your analogy with SUVs :-)

I do think throwing NonStops into Universities would be a good idea. It gets the name out there for a start, and that helps to solve the "I've never heard of NonStop" issue. However, this is a very long term investment - it takes a fair while for those computer science recruits to be promoted up the ranks to decision makers.

Richard Buckle said...

Dave, when I wrote "Consider IBM’s System z where ... a major push into universities, but when it comes to deploying “modern applications” it’s rarely considered, other than in combination with zLinux"

I was drawing attention to how IBM spends a lot with Universities but it's having little effect (in further developing the mainframe marketplace). It is a longer term commitment for sure and I see value - but I am yet to be convinced that providing servers to them helps gorw their market presence. These days, I would conjecture, decisions are nearly always driven by software, and the software stack options, moreso than any predisposition towards a platform and for NonStop to increase it's market presence needs to better support many "much-needed" infrastructure building blocks.

Make sense?