Poll question: follow-up question to HP-UX upgrade or cold install

[marie-noelle jeanson_1]

Hi,

For those who may have missed my follow-up question thrown in the middle of my prior thread which you may not have seen, the question was:

Let's say HP-UX introduces a new feature on the new OS (agile addressing, not to name it) that requires pretty significant changes in the system config and the system management:
would you choose to continue managing your system the old way or would you jump in and use the new way?
Which way would you like HP-UX to set by default after cold installing the new OS: Legacy or new?
(knowing that you need to learn a bunch of stuff to get up to speed on the new one, but it provides features nice to have. And HP will provide you with documentation on how to go from one to the other).

Thanks,

Marie.

[Arunvijai_4]

Hi Marie,

Q1) Would you choose to continue managing your system the old way or would you jump in and use the new way?

Need to look which one fits for me. (I will vote 70% old and 30% new)

Q2) Which way would you like HP-UX to set by default after cold installing the new OS: Legacy or new?

Mixture of both..

-Arun

[A. Clay Stephenson]

1) I would immediately try the new "improved" method on one of my Sandboxes.

2) Whether or not I then used it for real would be determined by Step 1.

[Bill Hassell]

Based on the new revisions starting with 7.0, I would never configure a new production machine to use a new config and management system. SAM was supposed to be the end of command line and config file editing back in 8.0 but I did not consider it stable (reliable) enough for production for another 5-6 years. I would try the new system on a sandbox system and go through a checklist of possible hardware and software configs, patching, Ignite backup and restores, mirror disk configs and failure replacements, multiple LAN cards, etc.

I think it would be fairly risky to set the default to new style as it may create very difficult problems right out of the box. This forum can provide a lot of expertise but the help provided may have no meaning on a system with totally different system management tools.

Although it may be possible to document a switch between the two methods, I'm sure there will be config problems that will confuse one or the other methods, leading to a possibly broken system (especially networking). And just like Trusted versus shadow and standard password systems, new (untrained) sysadmins may not know what they are using leading to a lot of confused questions. So the new system should have easily identifiable items to make the distinction.

[Pete Randall]

Like Bill and Clay, I start by installing on a sand box (my workstation). If I can't manage to break it, I'll install on the development server, making sure I can fall back if need be, either through Ignite backup or a split mirror. Once the developers stop whining, I will then install on the production machine, again ensuring that I can fall back if need be.


Pete

[Deoncia Grayson_1]

would you choose to continue managing your system the old way or would you jump in and use the new way?

I would try out the new way on one of my test boxes or as the others stated, on my "sand box" before trying it out on one of my critical test boxes. It all depends on how this goes will determine if I will use it in the future.

Which way would you like HP-UX to set by default after cold installing the new OS: Legacy or new?
I would prefer it to be a mixture.

[Patrick Wallek]

Like all others, I would have to play with the new features first before making a decision on whether or not to use it in production.

For cold install - default to legacy or new? --- Why not give me a choice during the install? If I had to upgrade/install a new machine before testing the "new" methods, I would likely choose legacy.

[DCE]

Marie,

Like all of the others, I would first get up to speed on the new features on a sandbox system. That would then determine how fast I move to the "new" way. As with all changes, eventually the "new" way will eventually become the default way - it is just a matter of obtaining the knowledge, and letting the first couple of versions go by.

I would prefer that on new systems there be an inititial choice between the two ways - with the new being the default selection.

[Eric Antunes]

Hi Marie-noelle,

1- I would try it on the Apps test system, noting all necessary steps and time to implement it;

2- If this "agile addressing" would be an improvement for ALL types of environments (OLTP, Datawarehouse, Web server, etc...), it should be default, otherwise it should be asked to choose between the new and the old feature.

Best Regards,

Eric Antunes

[Rick Garland]

This is a job for the 'sandboxes'

I try to be up to date with the latest and greatest but this also depends on work flow - do you have time to research?

Which way I go would depend on the benefits I see, find, or hear about in using the new methodology.

[Steve Boniwell]

From experiences with a range of customer (end user) sites I would observe that most users will not implement any new functionality unless:

a - they are buying new kit that has it by default

b - they are pushed into it by chenges in application software

c - It offers an obvious cost or performance benefit

Most of the time I have trouble persuading them to apply critical patches. Generally they want to upgrade rather than re-install if at all possible and do not appreciate any sort of change of environment.

Other than that I concur with comments already made - "Try it before you buy it"

Cheers,
Steve

[John Collier]

Although I cannot really add to the entries above, I have to concur with the general consensus:

Test it, then implement if the benefits are obvious.

As for the default on new, I am still the paying customer. You had best give me the option.

[John Payne_2]

I agree with 'those who came before'. Test it out, see what happens. If it is something we REALLY need, I test it in some more places, make sure it works, and away we go. Now if it's something 'nice to have, but not necessary', I would give it a best effort type of thing, but may not get there.

HPUX 11.23 has some pretty nice features, but we do not have to be there today. It took me years to get us off HPUX 10.20, this past year I got us off HPUX 11.0 onto HPUX 11.11. There is nothing critical pushing me to HPUX 11.23 at this point, so it will wait until at least 2007 at the earliest. (Although we have several Itanium machines in production, and we are running it, 95% of the machines are HPUX 11.11)

As far as how to manage, it just depends on what the changes did for me. I am the sysadmin (they call me an engineer) for 75 machines. If the new management tools greatly simplify things for me, I would spend the time and effort to move to the new way, otherwise I stay where I am.

When I started using Ignite/UX, it was leaps and bounds better than cold install via CDrom. This past year I started using HP SIM to do some sysadmin tasks. (And schedule things so I didn't have to actually be there... :) ) Software Package Builder allows me to update things much easier also. There have been products I looked at that didn't impress me and I don't use.

Anyway, I don't know if this is any use to you, but there you have it.

John