UNIX Special: Shaken, Not Stirred

OREGON, US: It ain’t over till the fat lady sings. And in boxing battles, it ain’t over till the bell rings. There is no sign of this bell ringing for UNIX if Dan Olds, Principal Analyst, Gabriel Consulting Group says it right. But listening to Dan’s sharp and cogent arguments leaves not much room for doubt. For this exceptionally sharp, brutally-cogent and seasoned industry-watcher or analyst extraordinaire, not only is betting big on the rugged son-of-the-soil Rocky Balboa of the enterprise rings, but he can also dig out some subtle yet impressive layers of reasons that surprisingly go in favour of someone whom most have written off already.

So yes, some new beans as counted by Gabriel Consulting in a freshly brewed survey are steaming hot for UNIX, but it still takes Dan’s strong sensory powers to tell us just why rustic brawn is still far from the so-predicted sunset days, no matter how many new-generation challengers like LINUX or x86 stand up against the old champ. The match is not yet over. Industry can still watch some thrilling moments of pugilism. And as always, there will be blood. Dan helps us see where.

Your survey throws up new energy levels for UNIX. What makes UNIX still contemporary amongst views that it drove into sunset with the advent of LINUX, x86 and Windows OS etc? If yes, which genre has the most potential in the new era - Mac OS X Leopard/ IBM's z/OS/Solaris/HP-UX?

UNIX is still around because it has changed roles in the enterprise. As less expensive x86 alternatives (Linux, Windows) advanced from the bottom of the market moving upwards (starting with file/print servers, for example), UNIX systems have steadily added enterprise features that the x86 systems can’t match. These include vertical scalability (scaling a single application to many processors/cores), along with higher availability/reliability and advanced virtualization. At the same time, prices for UNIX capacity have come down as well. The modern commercial UNIX system has moved into the role of mission-critical infrastructure server.

So, who will grow at whose cost?

I think that the major operating systems will continue to survive, but the bulk of the growth will go towards Windows and Linux — which is natural, given that they are lower cost alternatives. I believe that commercial UNIX systems will continue to grow in terms of usage (bigger and more capable systems being shipped), albeit at a much slower rate. The mainframe will continue along, much as it has, primarily selling into their installed base.

UNIX is definitely facing some heavy challenges in a torrent of new alternatives? How many of these can be attributed to slower release cycles, UNIX wars, slow evolution of standards, latency issues, portability issues, flexibility, skills shortage, costs, server consolidation issues etc? Which ones do you opine as the big punches to focus on?

The key challenges for commercial UNIX vendors are two-pronged. First, proving (not just talking about) the value of their platform vs. x86 alternatives. They need to talk about throughput and workloads vs. what can be generated by other systems. They also need to discuss costs, and put forward a case showing that the higher acquisition cost of UNIX systems is more than counterbalanced by, for example, lower management costs, higher throughput, higher utilization rates, or higher availability.

Secondly, I think that anyone who knows Linux would feel fairly comfortable with commercial UNIX these days. This is something that the UNIX vendors need to explore and discuss with prospective customers.

How would you assess its relevance for mission critical apps or heavy workloads in coming years? For enterprises overhauling legacy systems in a major way? Do you see more migrations in imminent future?
{#PageBreak#}

In my research over the years, I’ve found that customers see their commercial UNIX systems as mission critical servers and that they see these systems as strategic in their organization. Most customers that have commercial UNIX systems don’t see themselves moving off the platform in the near or medium term.

Any observations on some concomitant technologies, hardware and other contextual pieces that might impact UNIX’s struggle in face of LINUX? How does its quintessential multi-tasking or pipelining characteristic intersect with the Cloud- or virtualization-generation of technologies?

In the survey, I asked a set of questions aimed at seeing how customers compare and contrast commercial UNIX vs. Linux. What I found is that with customers who have both operating systems, most of them see commercial UNIX ahead of Linux in terms of availability, reliability, virtualization, customer support and other data center-centric factors.

If UNIX indeed continues to stay alive in today’s pecking order, what would be (or should be) its advantages or differentiators? Does a strong and long evolution curve help it outscore LINUX in a major way?

I think I answered this question in some answers to other questions But, briefly, UNIX needs to continue to show how it’s a better platform for mission critical workloads — how it’s more highly available, provides more predictable performance, and is easier to manage vs. Linux.

In the UNIX universe, how would the orbits between IBM (Power line), HP and Solaris play out next? Would they continue investing in this space? And yes, would the price-discount advantage due to competition between the big UNIX three continue as x64 gains more steam?

Yes, the remaining three UNIX vendors will continue to be very aggressive when it comes to pricing. I don’t see any of them slacking off in terms of system and O/s development. This is a profitable business for all three and they won’t abandon it anytime soon, if ever.

Coming to the much-in-buzz Oracle-HP face-off over Itanium support. Would it affect HP-UX? Isn’t x86 gaining real development muscle over UNIX and is closing the gap fast?

I don’t think that Oracle dropping future version support for Itanium has had nearly the effect that Oracle hoped it would. I think that Oracle hoped it would be a death blow to HP-UX based systems — but I think it’s had an even more negative effect on Oracle, it’s caused a lot of mistrust with current Oracle customers. I think this move has backfired on them. The majority of development is centered on the x86 platform, of course. But there is a lot of commonality between Linux and UNIX, and porting a particular application isn’t all that difficult these days. We also have to understand that there are lots and lots of applications that will never and should never run on commercial UNIX. These commercial UNIX systems are data center mission critical systems, they won’t be called upon to run many of the workloads that run on x86 servers.

How do you interpret these numbers? Gartner finds out that- UNIX shipments went from 670,458 units shipped in 2006 to 437,414 units in 2009, with a slight up tick to 451,593 units anticipated in 2012.

Yes, shipped units are down and have been dropping steadily since the early 2000’s. But the systems that are shipping are larger configurations and offer a lot more performance for roughly the same street price. If you look at the performance of the systems that ship every year, you see that it’s way more than what is being decommissioned in customer data centers. It’s very hard to pin these numbers down, but I believe it’s true on the face of it.