The more things change

Linux and Windows have an unexpected commonality: development in both groups is constrained by the relative lack of change in Intel's x86 products - relative, that is, to hardware progress being made in the PPC, SPARC, and IBM/Cell communities where things like transactional memory, hardware scout, and parallel ultra-high speed buses are driving software developers to, and beyond, their limits.

One of the great things about working in IT is that there's new stuff all the time - and if some of the best stuff is actually old stuff that's new and improved, well I'm ok with that; provided only that it really is better.

So here's a question: is there anyone out there, really anyone at all who doesn't sell PC products or services, who genuinely believes that Vista and Server 2008 running on the latest and greatest from Intel are actually better - in any serious way than their XP/2003 predecessors?

The latest Linux products, in contrast, are obviously better than their predecessors - it's mostly with respect to installation procedures and oddball hardware support, but essentially every new distribution is genuinely preferable to what came before.

On the surface that's been true for Solaris and the BSDs too: each new point release has added some things and improved others, so you might easily think that grouping Windows versus Unix on the change dimension makes sense.

It does, but there's another split you have to take into account: a very big difference between what's happening with Linux and the other Unix variants.

This difference is that Linux developers turn out to have something very much in common with Windows developers: both are severely constrained by the lack of significant progress in the x86 hardware community. In contrast the people now working at Sun, Apple, and IBM to get the next generation PPC, IBM's Cell monobloc, and Sun's Rock processors into the market are facing a tsunami of hardware change that makes x86 change about as compelling as used beer in a rainstorm.

In some part this reflects an historical difference in focus: the original Microsoft desktop co-evolved with Intel's product focus on that market - and when Linux came along it got caught up in that same process: a "free unix for the 386" is just that: an x86 desktop focused product that can be used as a server. Sun and IBM, in contrast, have traditionally focused on servers and both PPC and SPARC development has been driven from both the server and dedicated device perspective, but not from the desktop one.

Since IT people outside data processing and Windows tend to be a lot less risk averse with respect to technology than users, this means that what we got is just what you'd expect: rapid adaptation on the server side and relatively little change on desktops - if you like car analogies think of Intel as a car being driven in reverse in competition with forward facing racers from Sun and IBM.

Today, however, the two change processes are so far out of balance that something has to give - and since I think it's going to be the traditional desktop that gets tossed aside, I think the real battle, the tidal wave of future change as it affects us IT grunts, is going to come from the demands wide spread acceptance of user-land devices like Apple's rumored 12" super iPhone and Sun's next generation Sun Rays will put on us.

The potential is amazing - you can imagine was well as I can, I think, what a 12" iPhone replacement for the traditional laptop could do for user productivity, but let me quote you (with minor edits for typos) a new Sun's guy's musings on getting his first home Sun Ray to suggest what that could mean for us if, or when, it starts to catch on more broadly:

Thursday Jul 10, 2008

Sun Ray Technology rocks !

So yesterday was a very interesting day. I received my Sun Ray @ Home thin client. Now a little bit of history. Sun has deployed Sun Rays throughout its offices. Very few employees use laptops. So for those of us who are mobile we use flexibility. When we need to go to the office we "hotel". We basically reserve an office much like a hotel room. Each office is equipped with a Sun Ray thin client.

Well I signed up for the Sun Ray @ Home program. A box showed up with a thin client connection box, router, ethernet cable and power strip. I already had a screen in my KVM setup so all I needed was the Sun Ray connection box. I hooked up the screen, ethernet and power to the box and in less than 30 minutes I was connected to my office terminal session. How cool is that. I can now work @ home and get access to my office apps that I would normally need to boot the laptop, sign in to the VPN.

The Sun Ray is responsive over good ole Comcast. Just like I was in the office or better. Yet it drawls a fraction of the power. Here's to the Sun Ray.

I like to think of this as a kind of second coming, because of course companies including NCD, Sun, and NCR/AT&T had earlier implementations of the same core ideas working in the 1980s, but as the wintel thing becomes less and less innovative (and more and more expensive) the power of the ideas behind product sets like Sun Ray and iPhone are going to become increasingly important to us.

So, bottom line, if you're like me and barely starting to assimilate technologies like Sun's T2s and IBM's cell into real business processes: better dig in, because the server side of the revolution is just getting started and once the users go into catch up mode, we're just barely going to be hanging on for the ride.