Re: New system with Guested Lpar

[CRPence <CRPbottle@xxxxxxxxx>]

  To be clear, I did not intend to imply anything about a crash.  The 
'skinny' primary concept is not merely for crashes.  The concept is for 
both unscheduled and *scheduled* down time.  If the hosted partition is 
an effective production [or another type of environment] which should be 
able to continue active, even while other partition(s) are down, then 
instead of using a production primary and one hosted, there should 
probably be a skinny primary and two hosted.  The latter is what I tried 
to suggest was the more prudent implementation for that environment.  In 
that manner the first and second hosted partitions can go up & down as 
required without impacting the other.  But since the skinny primary 
would be used effectively by nobody and for effectively nothing, that 
best ensures the primary would rarely need to be powered down, 
irrespective of planned versus unplanned.  Using the former model, if 
the hosting partition must be IPLed to apply maintenance, the hosted 
partition is also unavailable for that long... and if both need to have 
the same maintenance, then the hosted partition is unavailable through 
both applications; scheduled maintenance on the first makes the second 
longer and/or unscheduled down time.

Regards, Chuck

Pete Massiello wrote:
> I can't remember the last time any of my machines or partitions crashed, so
> with that said I would run a guested partition, as my environments and all
> my customers machines hardly ever crash.  
>
> This is just my .02
>
> Pete
>
> CRPence wrote:
> > Wilt, Charles wrote:
> > > > pete massiello wrote:
> > > >
> > > > <<SNIP>>  gives people a great way to test out a new version of the
> > > > OS. One drawback is that the HostING partition becomes like a
.>>> Primary partition to the HostED partition.
> > > >
> > > That being the case, do you think it'd be wise to consider going back
> > > to the old recommendation that your primary (HostING) partition, not
> > > be doing any production work?  That the HostING partition's only job
> > > be to host the partitions where the real work is occurring>?
> >
> >    That would be prudent, *if* the hosted partitions will be used in a 
> > manner that mimics typical use of the more thoroughly separated [and 
> > more likely permanent] Logical Partitions.  However I would expect that 
> > most hosted /i/ partitions would typically be used in ways that will 
> > eliminate most concerns for the impact of an IPL of the primary.  That 
> > is, hosted partitions will probably be used for various testing, whereby 
> > each hosted partition is considered a _temporary_ [non-production] test 
> > environment.  A hosted partition is unlikely for instance, to be the 
> > target of continuous updates to maintain an up-to-date image of the 
> > primary or another hosted partition, in order to maintain a somewhat 
> > permanent testbed.  <<SNIP>>