I'd have to agree with Rob here.

We have a limited amount of CSE here.  Just xlink and only on particular 
volumes that contain a certain very large server that doesn't want to be down 
for more than reboot time and we do a lot of PORs.   That has its risks too 
without a true shared directory.

I wouldn't wish a full implementation on a newbie.  Nor a conversion later 
(which has got to be harder than a fresh install of CSE 2.0 (ps. I made that 
name up.. I know nothing ).  The full implementation involves understanding 
each and every one of your mindisks... What you can share and what you musn't 
share...  Example... EREP 191, DISKACNT 191, don't share those but you need one 
of each on each system that is read writable.   You can share RSCS 191 but you 
want different configs...  You may have to write fancy profile exec's to get 
what you want.

The upgrade path is an unknown.  You'd have to figure it's a z/VM upgrade 
likely.  Would you also have to convert off of CSE at the exact same time and 
onto the new thing?  Does the whole cluster need to go at once?  Are you even 
allowed to upgrade all of your systems at once?

I would think that the future thing would still require you to learn what each 
disk is for, but perhaps with a better way of telling the directory what you 
want it to be known as on which system, and avoiding some of the special 
coding.  Hopefully it won't also make you squeeze 65,519 cyl minidisks into 
65,510 cylinders.   That'd be a lot of fun...

We had 2 systems fully implemented but took it apart.  We decided the risk of 
outages due to a cse systems programmer error wasn't worth it.

I'm sure Jim E will advise Sunny correctly with his insider knowledge.  If not, 
Rob can beat him up later.

Marcy 

"This message may contain confidential and/or privileged information. If you 
are not the addressee or authorized to receive this for the addressee, you must 
not use, copy, disclose, or take any action based on this message or any 
information herein. If you have received this message in error, please advise 
the sender immediately by reply e-mail and delete this message. Thank you for 
your cooperation."


-----Original Message-----
From: The IBM z/VM Operating System [mailto:[email protected]] On Behalf 
Of Rob van der Heij
Sent: Wednesday, July 29, 2009 2:48 PM
To: [email protected]
Subject: Re: [IBMVM] Where is z/VM CSE

On Wed, Jul 29, 2009 at 10:50 PM, Alan Altmark<[email protected]> wrote:

> The things you learn while implementing CSE (any or all of it) will prove
> invaluable in the transition to the "single system image" of the future:

So you're really suggesting people to start doing CSE to get up to
speed for a future z/VM release? Like getting PVM and RSCS installed?
I see that some people started to migrate their Linux data to new
volumes so that they could reserve the extra cylinders for the CSE
tracks (beyond 3390-9).  Or move around system data to share certain
objects in the collection, dealing with the problem that default
installation breaks the CRR server in CS (and that the examples on how
to do it are wrong too).

Sure, I see there's a lot of high-level and conceptual things that you
need to address in either case. But I very much doubt you develop such
insight by trying to implement CSE with current components. Maybe
different when IBM would already provide guidance explaining what
things will be different and how you migrate.

Rob

Reply via email to