In a move to “simplify” pricing and create hardware “parity,” Oracle has announced once again that they have changed their database pricing for the second time in less than six months.
After listening to Jacqueline Woods, Oracles vice president of pricing and licensing on a conference call explaining the latest policy shift, it seems clear that Oracle has achieved neither simplicity nor parity.
For example, Ms. Woods explained that the new policy assigns pricing factors of either .25 (Sun T1), .5 (AMD and Intel), or .75 for all other multicore processors.
This means that to calculate how much a processor license costs if you were buying a server with the new Sun UtraSparc T1 with 8 cores you would calculate as follows 8*.25*40,000=$80,000 per socket to license Oracles Enterprise Edition database.
If you were to purchase Oracle on a AMD dual-core chip, you would calculate the license costs as 2*.5*40,000=$40,000 per socket.
Do you get it? You better start pulling out the spreadsheets now to start calculating that new purchase, especially if your shop runs different types of servers.
How exactly does this change make things easier for customers? I get that Oracle is in a bad place right now. I mean, they have essentially lowered their prices for the second time in six months and yet I doubt many will be all that excited about the change.
Its just too difficult to accurately maintain such a convoluted approach. Dont forget, many customers still have older versions of Oracle licensing models gone bad that they still must maintain or convert to the new model at a rate that is always much too favorable for Oracle.
Perhaps its just time that Oracle admitted it has no clue how it should license their software in the face of a changing technology landscape.
Customers hate change, especially when it comes to licensing, which has traditionally been the number one sore point that users have had with Oracle.
This will not stop here, that you can count on. What happens when we get 10 cores or 20? Oracle cannot continue to change their policy every six months. If they do, the market will abandon them.
People hate too much change, so Oracle better figure out a viable long-term plan pretty quickly. Multicore chips are mostly about power savings and cooler operating temperatures than they are about performance.
Ms. Woods claimed that dual-core delivers between 1.5 and 1.75 the performance of a single core processor. Then what does an eight core processor deliver? These performance numbers are also highly dependent on the workloads the server is expected to execute, so the whole argument is spurious at best.
I believe that what the market wants first and foremost is a simplified pricing scheme. Oracles new plan is a poor example of that.
As I stated six months ago when Oracle announced its last pricing change, the licensing model must be changed completely. Sell a single edition of the software with a per server based price. Then sell support packages which can be based on the level of service selected and perhaps the company size.
Of course, there are always options that can be sold as well, also on a flat per server model. Would such a model impact Oracles revenue? Certainly at first, but over the longer term the stability and simplicity could make up for the lost revenue with greater sales volumes and more profitable support offerings.
One thing is for certain, complexity will only erode market share over time.
Charles Garry is an independent industry analyst based in Simsbury, Conn. He is a former vice president with META Groups Technology Research Services. He can be reached at cegarry@yahoo.com.