Enterprise Unix Roundup: R.I.P, IRIX
It was, perhaps, inevitable. But it is sad, nonetheless. Another one bites the dust. Is the end of IRIX the result of poor business strategy? Or is it the inevitable conclusion of a Unix platform going head-to-head with Linux?
On September 6, SGI sounded the death knell for yet another Unix operating system/platform combo: the IRIX on MIPS platform. By the end of 2006, all production on IRIX will cease, and in March 2007, SGI will no longer take orders for the product.
To say this was a surprise would be an error. Industry watchers, analysts and humble pundits saw this as the most likely outcome after SGI announced it would file for Chapter 11 bankruptcy in May a state that the company is still under. SGI's goal is to emerge from Chapter 11 around November, and it is very likely the trimming of IRIX from the product line is part of the company's reorganization plans.
But to be honest with you, I started wondering about the future of IRIX when SGI first announced its Altix line back in January 2003. At the time, the news was hot. There was much FUD in the air about how Linux did not scale well past eight processors. SGI's announcement of a new line of Linux servers that could physically scale up to 12 processors with 96 GB of memory, plus a supercluster product line that scaled up to 64 Itanium 2 processors and 512 GB of memory in a single node, was to put it mildly a bombshell.
Amid all the hoopla about the new Altix Linux/Itanium line, a complete departure from SGI's homegrown IRIX/MIPS products, it was hard not to pose the question even then: "what about IRIX?" SGI seemed to have its bases covered. The IRIX products were still in demand for customers' channels specifically, the high-end scientific and graphic industries, where they would remain in demand. Of course, that didn't jibe too well with its other statement: The commodity Altix machines would also be sold to the same customers.
It was hard to criticize SGI's choice to sell commodity and proprietary products to the same customers, especially given how likable the company can be. It may seem a bit casual to base an opinion about a company on that, but my gut reaction was that this company's heart was in the right place, and it was determined to service its customer niche, no matter what. The fact that SGI will continue support for IRIX until 2013 is further evidence of that.
In my many conversations with SGI, I always seemed to come back to the same question. In terms of performance, the Altix line kicked some serous butt. Why not expand it into other channels that could also use those performance levels? There was some talk about that, but to move into those channels meant moving into areas where HP, Sun Microsystems and IBM dominated. In the high-end technical channels, SGI ran the roost, with only limited competition from the other three major Unix vendors.
In other words, SGI was content to play it safe.
To give credit where credit is due, SGI did take risks. The Altix line itself demonstrates that. But to sell commodity-based products in such a tight market, it was little wonder IRIX sales fell. SGI was, for all its good intentions, cutting its own throat. Maybe it was counting on Altix to get its foot in customers' doors, whereupon it would be more free to sell IRIX/MIPS machines. I'm not sure what its plans were. Whatever they are now, it's clear SGI will be doing it without IRIX.
IRIX has now followed in the path of Tru64, Reliant and DG/UX into the Unix graveyard. Hopefully, this will not mean the end of SGI as well. Altix is still an excellent product line, and HP does not have a commodity offering that can shift into SGI's specialized channels any time soon. HP-UX is moving from the PA-RISC platform to Itanium, but it may not be enough, since SGI has a big head start in porting IRIX apps to Linux. Altix servers are still, as SGI loves to point out, breaking performance records left and right.
On a grander scale, it is hard not to pose the larger question: Is this simply going to be the way of things for proprietary Unix operating systems that try to go head to head with Linux? Is Solaris, or HP-UX, or AIX going to be the next termination we hear about?
It's possible. AIX has a big user base and IBM is a powerful support entity, and it could balance AIX and Linux sales. Sun may have saved Solaris by making it open source, assuring it will survive in some fashion in the years to come. As for HP-UX, there you have a problem. Squeezed by IBM and Linux, HP's Unix product, I believe, is in the most danger of getting the axe. The recent scandal on the HP Board of Directors (and am I the only one who finds great irony that one of HP's server lines is called "Integrity"?) isn't helping things, either.
While we ponder that future, let's not forget about IRIX and the contributions it has made for science and technology.
Goodbye, IRIX. It was nice knowing you.