[Info-vax] Message from HP.

[johnwallace4 at yahoo.co.uk]

On Tuesday, 10 December 2013 13:13:25 UTC, John Reagan  wrote:
> On Monday, December 9, 2013 2:20:23 PM UTC-5, johnwa... at yahoo.co.uk wrote:
> 
> > On Monday, 9 December 2013 17:20:47 UTC, John Reagan  wrote:
> 
> > 
> 
> > > On Monday, December 9, 2013 11:59:00 AM UTC-5, Gérard Calliet (pia-sofer) wrote:
> 
> 
> 
> 
> 
> > 
> 
> > Try this for size, in the absence of the Hal Massey presentation:
> 
> > http://ftp.hp.com/pub/nonstop/ccc/nov0608.pdf
> 
> > 
> 
> 
> 
> Read a little farther down that same .pdf and you'll see that LSUs are no longer used.  They exist on H-series Itaniums but they don't exist on J-series systems (ie, blades).  It is now software-only.  I suspect that NSK on x86 will be a software-only solution.

Maybe so. There were some bits in the Illuminata paper that struck me as (at
best) insufficiently detailed, but in a vendor sponsored 6 page paper that's 
not really a surprise. 

For example, there's a tease about the future possibility of resurrecting
triple modular redundancy which had historically been an option of interest
to 10-15% of the NonStop market. The blade-based NonStop couldn't do it (at
the time of the paper).

The role of a Cluster IO Module (CLIM) is somewhat skipped over as well
(standard redundantly-configured Proliant server+connectivity, software details
totally unclear).

Regardless, if the NonStop folks are going to be confident enough that suitably 
qualified x86-64 hardware suits them, and that all they need on top of that is a
decent OS of their own, can the VMS world "leverage the synergies" (yuk) of 
e.g. having some other organisation already having done lots of hardware 
qualification work? (That should have been an option long ago anyway to anyone
who wasn't inseparably attached to IA64).

I'm thinking one difference between the VMS world and the NonStop world is the customers.

The finance sector (and in particular the trading operations) have been looking 
at significant increases in volume, and therefore have needed to buy more kit 
more frequently (be they NonStop customers, VMS customers, or more recently, 
Linux customers).

The VMS users that Gerard mentioned run (ran?) factories, maybe utilities such 
as power and water (both, markets I used to know). These customers have 
probably not needed that much increase in power in the last decade or more. So
the manufacturing customers haven't needed to spend much money with HP (or 
CPQ).

But if these folk want to move off VMS, they're looking at spending quite a lot 
of money. Not just a few x86 servers and a few Windows licences plus some 
VMS-emulation middleware (or VAX/Alpha emulation). In the absence of a visible
future for VMS, these folks will be looking at a redesign from scratch, 
preferably on an OS/supplier combination with a ten year future. Can't do that
with Ruby on Rails and a few agile programmers.

You can maybe do *some* of it with a smart(er) ERP system and a redistribution 
of workload between IT kit and PLC kit. Jan-Erik (?) mentioned this - surely 
there's nothing that SAP can't do, given a big enough budget?

But how workable/affordable will that be in (say) the automotive sector where 
there is years of business logic and factory floor knowledge in the VMS applications?

Even though there's been relatively little visible spend for HPQ on these 
setups in the relatively recent past (why would there be?), there's going to 
have to be money on the table here somewhere, either to keep these setups going 
on VMS, or to move them off VMS. 

How much money, to do what, and who is going to get it, may be questions of 
interest to some folk.