PA-RISC and HP-UX support life

[Charles Holland]

OK this is not the first time the subject has came up. Most recently in

http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1017524

Let us try and seperate the 2 worms...

If given the multiple statements that we should be thinking about moving to the Itanium platform because the 8900 series chips will be the last PA-RISC chips produces.

Both HP-UX 11i v1 and v2 have a end of support date of 12/31/2013. Version 11i v3 is due out this year, but we don't know the end of support date yet.

HP is going to discontinue the PA-RISC but NOT do away with HP-UX.

So if you were thinking 8-10 years down the line you would be thinking HP-UX 11i v(insert # here) on Itanium... or it might be HP-UX 12....
Or go to Linux of your choice, or (gag..choke) Window$ because once on the Itanium you can run any of the three.

Clarification please if I am incorrect.

[Matti_Kurkela · Accepted solution]

I think you're absolutely correct.

About the HP-UX/Linux/Windows choice: very likely the applications you need will determine which way you will go.

The most important question is, can you get the application support level you want on the platform of your choice? On enterprise applications, the availability of support may dictate your choice of platform.

If your application is Java-based, you will not care much about the details of the platform you're running on, as long as you can get a reliable JVM with good performance for it.

The other programming languages may present more challenges:

If your application is 32-bit code, sooner or later the source code needs to be checked for 64-bit correctness, so it will compile correctly on 64-bit architectures. This may be somewhat painful, but it probably can't be avoided. If there is already a 64-bit version of your application, that's good.

To port an application from HP-UX/PA-RISC to HP-UX/Itanium, the application needs to be recompiled. However, since the APIs won't change at all, the porting should not require any major changes and very little minor ones.

To port from HP-UX to Linux might take somewhat more effort: if your application is hardware-related, some APIs may be very different, requiring re-implementation of some parts and some serious testing. If the application just "does things with files and TCP/IP", it might be almost as simple as the port from HP-UX/PA-RISC to HP-UX/Itanium.

The port from HP-UX to Windows would probably take the most work.

If you have the source code of your application, you can make the choice as you see fit. If your application comes as a pre-made binary from some vendor, the vendor's choices will probably more or less dictate your choices.

[Charles Holland]

Session closed

Thanks