Interesting. Why did they get rid of it? "Moore’s law could quickly make this type of licensing financially unattractive to the licensees. At the rate in which compute performance increases, the cost of a UPU license doubles about every two years. The industry quickly caught on to this and called Oracle out, which is why the UPU licensing model was very short lived." Well, clock speeds are no longer changing very much, so today I don't think there would be that much difference between "price = cores * speed" and "price = cores", so I suspect UPUs would be fine.
Currently, "price = cores" gives this weird market distortion where there's an incentive to get individually-fast, disproportionately-expensive cores to run software that might well be embarrassingly parallel. Perhaps it functions as price discrimination for those willing to put in the effort to customize systems like that. Eh, who knows.
These days, there are more viable alternatives to Oracle than there were back then. UPU was complex to administer, and had the primary purpose of extracting more money from customers. I suspect Oracle doesn't want to give customers a reason to start looking for alternatives.
I'm not sure there are really viable alternatives. If all you need is a big CRUD database, sure there are alternatives, but there always were. Anyone buying Oracle for that was throwing money away. If your enterprise is more fully committed to the Oracle platform, or if you need a support model where you can open a ticket had have people working on it 24x7, possibly on-site, until it's resolved, then no, you can't just drop in PostGres or EnterpriseDB.
Teradata a competitor at the high end and Microsoft SQL server is an example at the low customers still want significant support. The enterprise market is not competitive the way gas stations are, but major vendors do compete.
In many ways Oracle is the low end of these solutions, but few companies actually need giant databases with thousands of drives.
