[Info-vax] IBM Layoffs

[Kerry Main]

> -----Original Message-----
> From: Info-vax [mailto:info-vax-bounces at info-vax.com] On Behalf Of
> Stephen Hoffman
> Sent: 27-Jan-15 11:15 AM
> To: info-vax at info-vax.com
> Subject: Re: [New Info-vax] IBM Layoffs
> 
> On 2015-01-27 15:41:49 +0000, D W said:
> 
> > On 1/26/2015 11:39 AM, Stephen Hoffman via Info-vax wrote:
> >> Hosting VMS, for VMS customers.  This opportunity is pretty obvious
> for
> >> a vendor with licensing rights to VMS and related products,
> >> particularly if the vendor can create and manage pairs of data centers
> >> located within range of clustering and HBVS, and preferably on
> separate
> >> electric and communications grids and drainages.
> >
> > You can do most of that in one building. The Colocation site we use
> > gets power from two separate grids unlikely to be heavily loaded at the
> > same time (Sports stadium on either side) and gets data from three
> > telco trunks. I can't imagine what it would take to flood that area-
> > it'd probably have to be biblical.
> 
> Yeah, but when the {insert nasty product} being {manufactured, used,
> stored, transported, intentionally released} upwind goes {insert nasty
> event}, and you can't get into the location for a week?
> 
> Power grids, too, have suffered cascading failures.
> 
> Murphy, after all, was an optimist.
> 
> >> Irrelevant? Sort of.   Also sort of important, too.  Customers with
> >> applications operating on VMS today are probably going to want
> those to
> >> be run on VMS servers, whether emulated or actual servers.  For VSI,
> >> it'd probably be easiest to host Itanium guests, at least initially.
> >> Maybe eventually on x86-64 guests, once VSI gets the VMS code
> ported
> >> and then once somebody gets the customer applications ported.
> >
> > A VM would be great. Something like FreeBSD's jail(8) (Did any of the
> > Galaxy stuff survive the Alphacide? but that's more HW based than I'm
> > thinking) would not only do the trick for providers, it'd be nice to
> > have in general. Take several smaller customers wanting clustering and
> > let them each have slices of different machines.
> 
> Galaxy is largely firmware based, and Galaxy could — if somebody were
> inclined to heavily customize the ACPI support — be deployed with
> customized UEFI code.   OpenVMS has picked up the hardware
> configuration from the console, and the Galaxy firmware provided a
> coordinated form of subsetting.    Customized or extended UEFI is not
> trivial and would tend to be server-model-specific, but it does remain
> possible.  But no, Galaxy didn't make it to Itanium, and the same
> general difficulties that encountered would also apply to x86-64.
> 

Let's not forget that back then the HW vendor (Intel) did not have a 
virtualization architecture embedded as part of the hardware arch 
design like they do now.

As an example:
http://www.tinyurl.com/intel-virt-1   
http://www.tinyurl.com/intel-virt-2   
http://www.tinyurl.com/intel-virt-3   (good article)

Key question - 

Could Galaxy be adapted to use this new Intel HW virtualization
Architecture?


Regards,

Kerry Main
Back to the Future IT Inc.
 .. Learning from the past to plan the future

Kerry dot main at backtothefutureit dot com