Should I feel sad for Oracle?
Naah.
Does Oracle have any friends?
Naah.
Just pay up!
Oracle is set to appeal a decision which, if it stands, will require Big Red to make a US$3 billion contribution to HPE's top line. HP launched the action in Oracle's 2011 decision to quit developing software for Intel's Itanium processors. Oracle believed there was no future in the Itanium architecture, but HP pressed ahead …
Meh, Oracle probably cost you a year or two... but Itanic was on its way out before Oracle dropped support. MSFT dropped support a year or two earlier. Really all this meant for customers was that people needed to get new gear prior to upgrading to 12c, which they probably would have done anyway. Having written that, Oracle clearly did this to push their own gear instead of HP. Oracle should have just provided half-baked support for Itanic for another few years until it went away of its own volition. They basically murdered a chip on its deathbed.
Back in the days when many CPU types were common I used to regularly port software to new hardware, no big deal. OK: I needed to get the source in good shape (proper typedefs, etc) & once I ensured that I had got rid of all warnings when compiled with the Most Fussy option. All assuming decent compilers and a known operating system - which HP have got.
So why does Oracle, apparently, have a problem ? Their database already runs on several different platforms, so it can't be that hard. Agreed: every extra platform does involve more work, but it isn't as if they don't charge enough to cover that.
Porting software to HP-UX by the time the spat started was definitely no big deal. You sometimes needed to throw a couple of socket options due to its libraries supporting extra layer of security on sockets so any app could get a "firewall like" functionality. A few includes were in weird places. But that was about it. The rest was plain sailing.
So as far as technical argument for not having it running on one more platform Oracle has none. Whatsoever. It is all commercial and I am surprised it is not a subject to a competition authority investigation.
So as far as technical argument for not having it running on one more platform Oracle has none. Whatsoever. It is all commercial and I am surprised it is not a subject to a competition authority investigation.
This was obviously on the cards the moment that Oracle agreed to buy Sun's HW business. Until then HP-UX was the platform for most large Oracle Databases and the two companies frequently worked together. Once Oracle acquired a HW business HP suddenly stopped be "best buddy" (or most profitable business partner if you want a more realistic interpretation) and became chief competitor in the mistaken belief that all Oracle customers who just happened to run on HP HW would immediately switch to running on Sun/Oracle HW.
For a database engine it's a bit more complicated. Developers have to have a really good grasp of OS innards like memory and I/O management. After years of development on a specific OS you're usually looking at a codebase that is wildly different from others.
But yes, no disagreement about Oracle - dropping HP-UX was just a pure unadulterated nastiness.
/edit: Just what EveryTime says below. Good comment./
Porting the underlying database code is significantly more difficult than porting a general application. Locking, consistency, caching at every level, and quirks with every I/O system needs to be taken into account.
However the porting work was already done. Keeping things working still takes people and money, but the risk is much lower. Companies might grumble about the maintainers looking "retired in place" and only putting in easy 8 hour days. That's sometimes what experience looks like.
The real issue here was a management decision to renege on the contract. There was a written contract. External companies relied on the contract. The business might not have been as profitable as expected. That risk is exactly why there was a contract.
Oracle has a reputation for "that is just paper, this is now the reality". That usually works for them. You aren't going to sue them when the project is 80% complete and your job is on the line. Or when your company's vital information is locked up in their database. Or you need them to support the next generation widget you are about to introduce. You look at reality, think of the line "pray I don't alter it any further" said by Larry with a Vader voice, and bend over again.