Jul 15 ’09
Linux on System z: The Sun Saga Continues
It appears the proposed IBM acquisition of Sun isn’t going to happen after all. That’s both positive and negative. The Linux world has both opportunities and challenges in either case, and maybe Oracle has an idea that may take them places they never thought they could go.
First, the positives. For Sun, it certainly saves them the ignominy of being consumed by their worst bogeyman. The pattern of pushing the “reconsider the mainframe” message, and the TCO studies of discrete hardware vs. the traditional mainframe-plus z/OS combination, were well-thought out, and the idea of having to admit they couldn’t make it work in the end would have been bitter, indeed. The Sun folks have built fairly sensible hardware, and through some really smart and strategic investments in the early ’80s and ’90s, the company managed to produce a generation of IT people familiar with their offerings. Oracle is a known entity; it has strong historical affinity with SPARC hardware, and Solaris is still by far the most popular traditional UNIX operating system available today. That’s ample reason for a customer who’s supporting existing Oracle installations to continue with what they know. As I’ve said before in this column, it’s not technology that costs money; it’s how you interface technology with people.
IBM wasn’t able to close the deal; to some extent, tweaking the price at the last moment was a cruel strategy. As Gen. Grant demonstrated, if you have to lose to your worst enemy, it’s important to allow the loser to save face in some way. An IBM/Sun merger would have been, in my opinion, a cultural match much better suited to producing interesting stuff, and IBM’s recent thoughtful policies related to intellectual property releases could have addressed a number of issues in Sun’s relationship with the open source community.
That said, there are some things in this deal that strike me as particularly disturbing, both for other hardware vendors (not just IBM, but also Dell and HP) and for customers at large. First, from my own experience with implementing Oracle applications, the hardware is by far the smallest part of the problem. Configuration, setup, and deployment have always been the hardest parts of getting an Oracle application to run, and Oracle currently has no way to address these issues. Most Oracle customers adapt their business process to what the software can easily do, rather than the other way around, which isn’t right. Oracle could combine hardware and software into a single turnkey appliance package that contains ready-to-go, Oracle-branded hardware and software (just add power and network). Most of the tools to make this easy are already there in Solaris (and in Linux, for that matter).
Second, if we consider the relationship that Oracle and IBM have had for a while on System z, they do a lot of deals together, but it’s always chafed Oracle to conform to IBM programming and pricing guidelines. Much of the financial case for customers doing Oracle on Linux on System z has revolved around using per-engine pricing for Oracle software on Integrated Facility for Linux (IFL) based on IBM’s suggested pricing model for IFL and Linux software: Cramming a lot more database instances and users onto existing systems that don’t increase floor space, environmentals, or your z/OS software bill is a highly compelling case. With this acquisition, Oracle now possesses an enterprise-class operating system of its own: Solaris. Adding a small investment to what Sine Nomine has already done with getting OpenSolaris running on System z hardware would permit Oracle to control the entire software stack on System z from the metal up—a first for Oracle—with no requirement to comply with IBM guidelines for pricing z/OS or z/VM software (because LPAR is a native feature of the machine and neither IBM operating system is required any longer). The investment could easily be recouped through licensing costs for Oracle software. In addition, this would mean catching IBM neatly in the Seven Dwarfs trap of not being able to block outside operating systems that comply with published interfaces.
Third, I’m concerned for the talent pool that remains inside Sun and its ability to continue to innovate. This merger appears to me to be similar to the DEC/Compaq/HP mergers—different cultures, different goals, and different ways of doing things. It’s anyone’s guess how long they can hang onto the talent they had at Sun—the number of ex-Sun résumés in online databases is already rapidly growing.
So, the Linux on z landscape radically changes as we watch this go down. If Oracle plays its card, we can expect to see some very interesting changes in the players.