Original URL: http://www.theregister.co.uk/2009/03/19/ibm_sun_deal_comment/

Sun and IBM - What price Bigger Indigo?

The server world goes mad

By Timothy Prickett Morgan

Posted in Servers, 19th March 2009 03:21 GMT

Comment Innovation loves a crisis, Sun boss Jonathan Schwartz likes to say. But perhaps not so much as your competitors love your particular crisis. Witness the way IBM has quietly but gleefully watched Sun's power and fortunes wane in the aftermath of the dot.com bust.

Every one of those wisecracks by Scott McNealy and Ed Zander a decade ago about "the world versus IBM Global Services" may come back to haunt Sun's founders and employees if the rumored takeover deal of Sun by Big Blue does indeed come to pass.

As of the market close on Wednesday, no deal had been announced. But we hadn't seen a firm denial from Sun or IBM either. That, plus reports confirming the original report in the Wall Street Journal coming out of the Associated Press, Reuters, and the Financial Times, would seem to indicate that Sun and IBM are indeed talking about some sort of takeover deal. If a deal is in the works, the companies should make an announcement before the market opens on Thursday, and if the rumor is not true, then someone should go to jail for pumping up Sun's stock by 79 per cent to $8.89 per share.

That IBM would shell out nearly $8bn to buy Sun is stunning to people who have been in the computer business for a while. Not just because that is the largest acquisition that IBM will have ever done in its history, not just because Sun can be acquired for such a small amount of cash, but because it seems so unnatural.

Sun and IBM are allergic to each other. IBM is the button-down Puritan from the East coast (no matter how hard is has tried to loosen up, it is still as stiff as a new wingtip). Sun is the young, wisecracking hipster from the West coast (well, formerly wisecracking until it was crushed by the dot.com bust). If this were a Hollywood love story, there would have been a chance meeting and lots of high jinx, and then a romance would have ensued. But in real life, when you marry someone not suited to you, one of you ends up with the goldmine and the other one gets the shaft.

For amusement sake, let's assume the rumors are true and that Sun has shopped itself around to Hewlett-Packard (which apparently rebuffed Sun's advances) and to Fujitsu (which apparently can't come up with the cash). Maybe other companies have been approached by Sun for a potential acquisition too.

Let's even assume, for the sake of a good story, that the entry of Cisco Systems into the server racket on Monday provided the impetus to put some fire into Sun's eyes. Perhaps the rapidly decelerating server market, Sun's financial results for the quarter ending in March, and IBM's bragging that it is Mister Moneybags in 2009, despite the down economy, have compelled Sun to gussy itself up a bit and place a call to Armonk, New York.

Personally, I find this image repugnant in so many different ways.

I happen to think the systems market needs more, not fewer, vendors, and the market needs different kinds of boxes, both virtual and physical. But you would expect a systems editor to say that, now wouldn't you? And as for doing what is necessary, I also know what that is like too. I shut down a big chunk of my IT Jungle site and came to work at a competitor because I could not match the scale of The Register.

You fight as long as you can, but you have to take your opportunities when they present themselves. You have to deal with what is, not with what you wished, and you can't look back. Maybe Schwartz is having that kind of moment.

So how might Sun and IBM fit together? Well, first and foremost, IBM always thinks of recurring revenue streams and steady customer bases. Sun has never broken out its Solaris business (licenses plus support) as a separate line of business, but this is the jewel of the company, every bit as much as the z/OS and related middleware stack on IBM mainframes are for Big Blue.

It is basically free money, so long as customers can't easily move their applications. Sun's strong presence in the financial services and telecommunications industries is also something that IBM has craved. So that is worth something too, if those customers want to stick with IBM/Sun.

Bigger Indigo

Still, with that, the server and storage portfolios of the supposed Bigger Indigo would be a mess. A bunch of architectural choices would have to be made, unless IBM just decided to put certain products into stasis mode and ride them down, like HP did with Tru64 Unix and AlphaServers. None of us on the outside of Sun really know what the status of the high-end "Rock" UltraSparc-RK processors is and their related "Supernova" servers, but their long-term longevity at Big Indigo would definitely be called into question.

To help cushion an acquisition deal, I can envision IBM selling off the Rock and Niagara server lines to Fujitsu and reselling Sparc boxes made by Fujitsu while porting Solaris to its own Power Systems and running Solaris inside logical partitions. (The "Polaris" project to do an OpenSolaris port onto Power does exist, but IBM and Sun have both been unwilling to provide any insight into the development effort despite repeated requests. Ditto for "Sirius," the port of OpenSolaris to mainframes). If Rock machines were to be further delayed, they could simply be written off and buried in the accounting for the acquisition. In this case, IBM would get what it wants: more volumes for its Power Systems.

A Sun deal would also give IBM something else: a Unix variant that it controls that runs on x64 iron. As far as anyone knows, IBM does not have an x64 port of its AIX Unix, just like HP has said again and again that it will not port its HP-UX to x64 platforms. It could end up being cheaper to stand behind Solaris on x64 than create AIX for x64 if Power System volumes decline. If IBM loses any of those game console deals with Sony, Microsoft, and Nintendo, it will be under intense pressure to ramp up Power chip volumes somewhere to cover the costs of the very pricey chip fab in East Fishkill.

I don't think that IBM would go the effort of merging AIX and Solaris, unless it is a lot easier than I think it is. That won't make IBM any more money than separate Solaris and AIX, and it will cost money to merge them and disrupt application portfolios. It is far more likely that IBM would turn its QuickTransit emulation software loose on old Sparc workloads and move them to AIX that way.

As for x64 servers, both Sun and IBM are in trouble of different sorts. IBM's x64 business has declined dramatically in the second half of 2008, and the company has not really provided any explanation other than to blame the economy. And while Sun's "Galaxy" line of x64 servers have gained share in recent quarters, Sun has not grown enough to be a contender in the volume x64 market. IBM's and Sun's server designs are radically different, and by many measures, Sun's rack servers are better than IBM's and IBM's blade servers are better than Sun's.

Sun doesn't have an x64 tower business, which has shut it out of the SMB space. Not that having them has helped IBM compete with HP, Dell, and Fujitsu down there among the SMBs. But a consolidated and rationalized x64 lineup from Bigger Indigo could be forced to make sense. I think Sun's blades would go, but maybe not. IBM's blades have been losing market share like crazy. And who can tell what business line managers will do?

What I think is key is that Sun has lots of expertise in networking, something that IBM is going to need to gird its loins to do battle with Cisco in the "system" space. IBM stopped doing networking years ago, ceding that market to others. But it may be time to jump in again, and Sun has enough skills to create some trouble. Look at that "Magnum" InfiniBand switch, the "Crossbow" virtualized NICs, and the integrated 10 Gigabit Ethernet ports on the Sparc T series servers.

It's a start, and there is no reason IBM and Sun could not cook up a unified fabric of their own. A couple of tiny acquisitions here and there to speed it up and - poof! - networking business. (Not that this always works. Look at all those storage companies that Sun bought over the years that didn't really do a damned thing for the company except raise hopes too high).

Then there is all that open source software and the street cred that Sun has attained (well, by its own estimation) through letting go of its software. If there is one thing that IBM has shown no propensity to do, it is to open source its own software. No surprises there - that is where the profits for the company come from. IBM will leverage and endorse other open source programs, and it is not shy about committing resources to open source projects, but Big Blue would never go as far as Sun has.

But you can't put those Java, MySQL, OpenSolaris, Java Enterprise System, NetBeans, OpenOffice, and other genies back into their bottles. You can bet that IBM would want to monetize them and would probably do a better job than Sun has. (IBM is still selling and supporting Informix databases, so it knows how to keep software it would normally have wanted to crush a-going). IBM certainly craves Sun's influence with academia and the development community too, but this is more about seeding future markets than making money today.

Whatever financial and consulting services businesses that Sun has would disappear into the gaping maw of Big Blue with nary a burp coming out.

And as for research and development, IBM is about eight times bigger than Sun, but has an R&D budget that is only about twice that of Sun, at $6bn. I think IBM, if it bought Sun, would come in and freeze a lot of projects and see what IP it could turn into money over short and long terms, much as former IBM chairman Lou Gerstner did when he took over IBM in April 1993.

IBM uses research and researchers to make money. Period. Sun seems to be having a lot more fun playing around in the hopes of making discoveries that lead to money. IBM has always been very disciplined about research, especially since the mainframe tanked in the early 1990s. Sun should have cut R&D a long time ago, and if it had, maybe Wall Street would not be so heavy on its back these days.

But it is hard to change your ways. That's why sometimes you have someone else come in and change them for you. Then again, if this merger talk is all hogwash, maybe you don't. ®