Can we drop the flame war for a minute

Let's cut the geeky flame wars and try and agree on some facts. IBM have a ton more money than Sun, and have ridden the Linux wave pretty well in the large accounts where they make their real money. Sun missed the boat and despite some astonishing innovation in their SPARC/Solaris heartland have not been able to catch up and convince people to stay with them instead of defecting to one of the other vendors.

Those are the facts. Now the opinion.

If you are IBM, what's in it for you? You might acquire Sun to get their customers, or their current or future technology/IP, or their people.

"Sun's customers", I think is not sufficient reason to pony up $6bn. Telcos and academics for example, both famously loyal Sun groups, can be picked up for free simply by waiting for Sun to die. (Academics can also be 'bribed' more cheaply: $1m worth of Linux blade servers to a college buys you an awful lot of goodwill.)

"Their people". Sun have some really, really good people...plus some idiots. I think IBM could pick up the Cantrells and Goslings for much less than $6bn. Bechtolsheim might be more expensive...but he would probably not jump anyway. But he might not say no to a few mill in seed money for his next startup. Cisco do this all the time; they fund a bunch of Cisco execs to leave the mother ship, build a startup and then buy it back.

"Their innovation". Very hard to put a price on this one. It is Sun's crowning achievement that they continue to innovate through thick and thin. Are DTrace and ZFS worth $6bn? They're certainly stellar technology. But $6bn? No. Unless, that is, the patents for them are so broadly granted that they would permit IBM to go after netapp, emc, hitachi and so on. I really don't think IBM would risk a patent war like that.

Conclusion? IBM should not buy Sun. If there are things in Sun that they want, they should destroy Sun in the marketplace and then pick up what they want for a song.

Food for thought #2: What about RedHat buying sun for their technology? Buy the business, sell the galaxy line to Dell, and put zfs and dtrace wholeheartedly into linux. That would be interesting indeed.