I'll disregard the idea that I have that you do not seem to be very 
knowledgeable with GDPS (or sysplexes for that matter) and its many modes that 
have been released this past decade and a half;

LPM would only do what you describe if the image you're trying to move is 
healthy and running fine, not so much if your image is down/crashed (disaster 
situation). At that point you have to boot the image on your other machine, 
there is nothing for LPM to mirror to your other machine if your original image 
is not running. You will still encounter downtime in this situation. So the key 
is; LPM is only interesting for planned outages of your hardware.

Also; while you're swapping with LPM, your applications will encounter slower 
response times, because LPM will have to synchronously copy working storage 
pages and disk frames when the OS requests them on the target machine. Your OS 
obviously has to wait while LPM satisfies the request, which might take a 
considerable amount of time on an electronics scale. So not only are you 
swapping fast when you could be swapping slow (planned downtime is that, 
planned, you should be planning, so a slow swap should be bearable), you're 
also inducing an impact on your business by slowing down your applications 
while you swap.

Considering you have a sysplex architecture, you're free to do what you want. 
Slow and steady re-ipl an LPAR while its work is offloaded to your other LPARS 
in the sysplex (planned outage). Or instant swaps (if you went that far with 
GDPS) for disaster recovery, or considering how fast GDPS can be, business 
disaster prevention while a physical disaster is taking place.

So when you acknowledge the limitations of LPM, that being when you can invoke 
it, and what it would entail in terms of response time impact, it becomes hard 
to come up with a case where you could actually use it.


_Jan


-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of IBMZOS
Sent: donderdag 28 mei 2015 2:31
To: [email protected]
Subject: Re: LPAR MOBILITY

Sorry but GDPS do not do this in seconds. it automate start / stop with 
approximatively 30 mn of interruption and do dasd swap in seconds (base swap of 
dasd).  So the minimum interrupt time is 30mn-1hour when LPM do this **in 
seconds**.

Sysplex do the job, but only in full parallel sysplex, which is very hard and 
long work.

I would like to have ONE Click on my Z/OS lpar to move it on another CPU, as is 
on Iseries with LPM.

The future is.. on Iseries :(

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to