In corporate IT, perception matters more than reality. When I was running a clustered SQL 2003 on Dual Xeon ( 4 CPU ) machines, actual throughput and transaction rate of Powerbuilder front app. Far surpassed Oracle 8 running on Sparc Solaris 8 with four CPUs. Now, this was an easy test to perform and prove, but the actual result did not matter. CIO heard good things about the Oracle on SUN, so I had to go with it. Just to show you how much this ignorance cost us, I had to pay Oracle license per CPU. Clustered, so failover CPU had to be paid in advance. My Disaster recovery site needed another clustered solution, so all of sudden, I had to pay four times my full licensing fee! Compared to this, SQL server did not count my failover clustered server. My DR site would never come up unless my main site is down, so I paid very little fee for loading it there. At the end of the day, Oracle solution was ten times the cost of SQL solution, but there was no reasoning it with my CIO.
This is typical of IBM’s way of doing things. They lease things. I remember my company paying thousands of dollars each month for leasing obsolete VSE running on old mainframe. We were running CICS so we had to pay! And I think this is why both Oracle and IBM jumped on SUN’s fire sale. It is not about JAVA! It is not about MySQL! It is the hardware business client list! These are fortune 500 companies that will in effect, lease out any solution if they can get them to use it.
JAVA is important. Write once, run everywhere sounded intoxicating at first. But, with success of AIR, Flash and Silverlight, JAVA is getting less relevant every day. Unless there is a unified OS + JAVA Sandbox platform, it is nothing more than an interpreted platform. It’s life span is questionable. Maybe mobile OS will be their last chance.
MySQL is run by most LAMP – Linux/Apache/MySQL/Php – shops. When I was building out a spam filtering solution, I kept my database in MySQL. It is fast and lightweight. However, it is only recently with version 5 that we began seeing features like triggers and transactions. When it comes to rock solid features like remote procedures and replications, no one will risk his career on MySQL database. It will always be poor-man’s oracle, like in stereo , Hafler system was called poor man’s McIntosh. This is not a revenue generating venture. Service side might bring a few dollars, but, this is nothing more than a bait and switch product that Oracle will use to sell their Oracle database.
To both IBM and Oracle, the idea of SUN SPARC turn key is absurd. Oracle and IBM, both have been selling Oracle 8, 9, 10 on intel and DB2 on x86 for years. With 64 bit OS and i7 CPU, Sparc will never catch up on performance. Besides, development cost of Sparc CPU is prohibitive. SUN saw the writing, and they went x86 for that reason. Therefore, the future is on X86 – 64bit.
What about the Solaris? Open Solaris? Well, the Solaris on Sparc is usable. Solaris on x86 was stillborn. With the new convoluted management system on Solaris 10, Solaris is destined to go open source. Matter of fact, they have! They dropped CDE and went GNOME! OpenSolaris is already out! So, what does SUN have that is worth anything?
Well, believe it or not, their storage is HUGE! Their SAN capable storage is the cream that everyone is after. This is where Oracle can come in and dominate. Oracle can give an enterprise storage solution that does SAN and iSCSI! Throw in a small i7 server running Windows 2010 and Oracle 11 database, you have a system that you can lease. And you can call it a cloud computing and lease out the space to enterprises!
And this is why IBM is sour. They saw an opportunity to become an enterprise storage solution provider. IBM selling their own storage solution never went anywhere. On the mainframe side, Shark storage was over priced, under performing system that was nothing but a stack of SCSI drives. Mainframe – System 36 running VM can carve out Linux slices and it was nice, but leasing it for thousands of dollars every month for a mere terabyte of data storage was too much. Besides, ESCON was useless. It was slower than 10BaseT. 3-4 Meg was all I got! This is why IBM wanted SUN storage.
Was it worth all that Billions of dollars? Only time will tell. I think it is a combination of its client list and its hardware business. Together as a whole, I can see myself paying 9 billion dollars.
