In the world of servers, the commodity Intel server is currently dominant, but not by as much as you might think. It accounts for around 50% of all revenue spent on servers. Despite predictions that the Unix server market would enter terminal decline, being overwhelmed by loosely coupled commodity Intel servers, Unix is still going strong. Unix servers, primarily running Solaris, HP-UX and AIX, still account for about 35% of all server revenues. In fact, in growth markets (China, India, Russia, Brazil, Korea, etc.) Unix has a larger share; 38% of the market compared to around 30% in North America and most of Europe.

A Tale of Three Chips

Now take a look at the graph below. It’s from IDC (with annotation by IBM) and it shows the market share movement of the big three Unix players from 1999 up to now. You could title it, the inexorable rise of IBM’s P Series. In the period from late 2000 to Q2 2009, IBM has more than doubled its market share to a level previously unequalled by any vendor. IBM is walking away with the market, while both Sun and HP shed market share in roughly equal measure.

Although it would be an oversimplification, you can look at this as the tale of 3 chips, Sun’s SPARC chip, the Intel/HP Itanium and IBM’s Power chip. Take a look at the recent record of benchmarks carried out by these vendors and the Power chip is so far ahead it’s embarrassing. Someone should stop the fight.

After the dot com collapse stripped Sun of a good deal of its revenues, it was always going to have a hard time funding the continued evolution of SPARC. In contrast, HP has no-one to blame but itself. Itanium looks very much like a failure of engineering. You either have to conclude that HP made the wrong technology bet or that IBM’s engineering team has performed out of its skin. Either way, as the graph so clearly indicates, HP has only just been able to keep pace with Sun and IBM has been picking the pockets of both vendors.

If you look at this graph, the logic of IBM’s bid for Sun was clear. It would have instantly put IBM in the catbird seat. SPARC would have been laid to rest, IBM would have migrated the Sun customer base and HP would have struggled to stay relevant. But Sun balked at IBM’s terms, and Oracle stepped in.

Will IBM get the Sun Customer Base Anyway?

Six months will probably have passed between Oracle’s bid for Sun being accepted and Oracle finally taking control, and that half-year will further weaken Sun’s market position. Even if Sun just follows the trend shown on the graph, it will lose a couple of percentage points in market share. But that isn’t what appears to be happening.

Oracle believed, or at least it claimed to, that Sun’s customer base had been losing confidence in Sun’s ability to survive and the Oracle acquisition would fix that. The facts on the ground suggest otherwise. First, if you ignore the IBM trajectory, Sun was doing fine in competition with HP. Secondly, now that the Oracle acquisition is in progress, IBM is reporting a hemorrhaging of Sun’s customer base to the P Series.

Of course, IBM would say that. Right now, that’s what I’d say if I were IBM. But there are good reasons to believe that at least some of Sun’s customer base is sending out RFQs. Let’s list them:

1. IBM has been Running a Sun-to-P Series Migration Campaign. IBM has been doing this for years. It has a sharp attraction campaign that tones down the cost of switching, using “Power rewards” (like air miles, but for data centers) and an out-of-the-box packaged migration capability. This is not so different to the IBM mainframe migration campaign that Sun used to run, but the point is that IBM was geared up and able to get aggressive when the Oracle acquisition happened. HP was not.
2. Sun’s Telecomms Customer Base is Particularly Vulnerable. Sun does a large amount of business in the Telco sector, a sector in which IBM has traditionally been weak. The Telcos used to avoid IBM because of IBM Global Networks, which they saw as a competitor. That stopped when (in 1998) IBM sold Global Networks to AT&T. Since then, IBM has been growing its Telco footprint very successfully. In this sector the Oracle acquisition hangs like a cloud over Sun. The Telcos invest too much in hardware to be in the least bit sentimental about any vendor. 
   
3. Oracle is not a hardware company. There’s good reason to be skeptical as to whether Oracle can suddenly become one. Just as Sun has “iron in its soul”, Oracle has software in its soul. Sun failed to run a software business. So does Oracle have what it takes to make hardware work?
4. Oracle is Invested in a “Commodity Intel” Strategy. With Oracle you pay for the  software licenses and the hardware comes cheap. That’s how it competes (quite successfully) with IBM. So how does Oracle square the circle and make Sun hardware cheap?
5. Commodity Intel Didn’t Work for Sun. Sun tried shifting its customer base onto commodity Intel servers running Solaris, but the customers, in the main, didn’t want to know. They preferred to stay with SPARC or move to IBM. There was some success with Oracle RAC, but not much. Outside of Oracle RAC, Unix customers see commodity Intel hardware as risky.
6. Sun is Losing Money. The mathematics is remorseless. Oracle cannot discount Sun hardware without incurring further loses on the iron. It could try raising software fees on P Series hardware, to try to fix the equation, but that would be seen by customers for what it would be; a price hike. In these recessionary times, a price hike is hard to sell.
7. There Are No Rabbits in the Hat. It would be wonderful if Sun/Oracle had some new breathtaking technology ready to introduce – but there’s no evidence that it does. Indeed, the evidence is that it doesn’t. In order to pull a rabbit out of the hat, there has to be a rabbit in the hat.

IBM has raised its P Series revenues by about $500m per year in the last three years – except in recent quarters, when the hardware market fell out of bed. Then P Series revenues simply shrank less quickly than those of HP and Sun. IBM was happily eroding Sun’s customer base anyway, at a rate of knots. Right now, there’s every reason to believe that this erosion is accelerating.

So while Oracle is acquiring Sun, IBM is acquiring the Sun customer base.

«- 1 2 View All -»