Solaris 9 vs. mainframe Linux

COMMENTARY--A new version of Solaris isn't the same kind of media bombshell as a new version of Windows, but Solaris 9 is an interesting product. Clearly Sun has been looking at and learning from the competition.

It's ironic, but you can smell a whiff of Microsoft OS strategy in Sun's design for Solaris 9, which integrates a large number of previously separate products into the operating system: An LDAP directory server, an application server, Solaris Volume Manager (formerly Solstice Disk Suite), Solaris Resource Manager, SunScreen 2.5 Firewall, the Forte development tools--now Sun ONE Studio Tools--and the list goes on. It's good for an operating system to have features I guess. (And in an unintended reference to the Microsoft remedy trial, Sun actually brags about all the "middleware" included with the new version.)

I haven't actually tested the new version, but the strategy seems like a good one to me. After all free (as in beer) and/or bundled services like these have been a useful strategy for Microsoft and Linux distributions. Windows 2000 and most-- perhaps all--Linux distributions, already come with an LDAP server and Kerberos support. Windows 2000 comes with application server services.

Incidentally, there's more to Solaris 9 than a bunch of bundled services. For instance, it implements buffer overflow protection (reminiscent of StackGuard, a product of Immunix).

Sun's bundled services will have a lot more credibility with customers than the mass of bundled free (as in beer and speech) utilities and services that accompany the typical Linux distribution. Many of these are top quality, but many are simply thrown in with the distribution because there's no cost to do so. But Sun can compete in all these areas while still, like Windows and Linux, allowing customers to use third-party add-ons if they prefer.

At the low end of the market, all the new features may make a difference. Low-end Solaris/Sparc servers--what they call Workgroup servers--can be pretty affordable. It's not hard to see how customers considering Linux and third-party solutions might be swayed by an all-Sun solution that didn't cost too much more.

Once you get into the medium range and high-end, Solaris is heart attack-expensive (click here for sticker shock.) Companies considering it are probably more concerned with operating costs more than with the initial price. The bundled services are certainly a good thing for these larger enterprises as well, but they have more complex issues to deal with.

For instance, for a few months now there has been a minor war of words going on between Sun and Linux advocates over IBM's mainframe Linux implementations and whether they're a good idea. At the core of the arguments is the issue of server consolidation, which is IBM's most public marketing point for mainframe Linux, and Solaris 9 steps into this battle as well. Solaris 9's Resource Manager helps administrators consolidate multiple Solaris servers into a smaller number of larger, more powerful, more expensive servers--same basic notion as the mainframe Linux notion of consolidating servers. I still have more questions than answers about this whole consolidation issue; after my first column on the topic I got a ton of e-mail from you expressing more confusion and skepticism than enthusiasm, but there is a case to be made for consolidation in the context of a broad review and redesign of network architecture. Like I said last time, simply moving your programs and data onto a smaller number of more powerful servers will save you some dusting in the machine room, but you'll have just as much management to do.

But all other things being equal, the Solaris consolidation approach has an advantage in that it is a plan for consolidating to a Solaris server, not an IBM mainframe running an operating system (zOS) with which your administrators may have little experience. This is why IBM basically sells these boxes only to shops with existing mainframe installations. Will Solaris 9 appeal as a consolidation platform for sites running other UNIX implementations? That's harder to say.

In years past the right answer would have been that it doesn't really matter which is a "better" product--as if there's an objective set of standards out there for such a decision. What mattered was that the market was big enough for the both of them and everyone will make money because there's enough customers with needs that are well-served by each approach. Unfortunately, as owners of SUNW know, we're still in a tech recession. One day server sales may perk back up to late 90s levels again and everyone will be happy. Solaris 9 is not designed to move into any other markets, so Sun will live by the server or die by the server.

What's your take on Solaris 9's consolidation abilities? TalkBack to us, or send e-mail to Larry.

Stay focused: Sign up for Tech Update Today, the daily e-mail newsletter for those who need to know.