[Info-vax] The Machine - Dows Jones newswire report

[Kerry Main]

> -----Original Message-----
> From: Info-vax [mailto:info-vax-bounces at rbnsn.com] On Behalf
> Of Arne Vajhøj via Info-vax
> Sent: 29-Nov-16 8:42 PM
> To: info-vax at rbnsn.com
> Cc: Arne Vajhøj <arne at vajhoej.dk>
> Subject: Re: [Info-vax] The Machine - Dows Jones newswire
> report
> 
> On 11/29/2016 5:33 AM, johnwallace4 at yahoo.co.uk wrote:
> > On Tuesday, 29 November 2016 03:46:35 UTC, Arne Vajhøj
> wrote:
> >> On 11/28/2016 7:03 PM, Stephen Hoffman wrote:
> >>> x86-64 server boxes are around for multiple decades, low-
> volume
> >>> hardware tends to be really expensive hardware, lower-
> volume
> >>> hardware or software and that requires specialized
> programming tools
> >>> or skills tends to be more expensive,
> >>
> >> And if time has shown one thing then it is that high volume
> and low
> >> cost solutions tend to win in IT.
> >
> > Maybe. It may depend on the criteria used to define "win".
> >
> > High profit tends to win, but isn't always high visibility, whereas
> > high volume is often by definition high visibility. Exception to
> "high
> > volume = high visibility": not many people realise how
> dependent they
> > have been on ARM system-on-chip designs in their hard drives,
> TVs,
> > routers, etc. That said...
> >
> > "High profit" may mean a little profit on a lot of systems (e.g.
> > generic x86 tin, "ARM everywhere", etc), stuff that Joe Public
> can
> > readily see, names familiar from the technology/fashion media.
> >
> > Or it may mean higher margins on a much smaller number of
> systems
> > (e.g. the niche stuff that gets used in high frequency financial
> > trading and where customers don't mind paying a fortune
> because the
> > extra millisecond they save gets them at the front of the queue
> for
> > million-dollar profits rather than second in the queue). Joe
> Public
> > may not see much of these, maybe systems sell by the
> hundreds not by
> > the millions, but **they win in the market in which they choose
> to
> > compete** (for now).
> 
> Try compare the high profit margin products like Alpha, Power,
> SPARC, Itanium with those low profit margin x86-64 thingies. How
> did it go?
> 
> Arne
> 

Part of what happened is that technology has increased so much on every platform that most applications today will run just fine on the mid-level server from ANY vendor.

Just look at why the X86 server sprawl happened.

As each server was green fielded and replaced on a one for one basis, the new server was cheaper and faster than the one it replaced. Hence the system utilization workload % dropped on each server by 20-35% each time the green fielding happened.

[funny story - before each green field, the Cust would have a bake-off and look at perf numbers from different vendors, then offer big RFP's to different vendors. When it was time to green field again, the servers in question were only running maybe 30-40% at peak times. Hence, how important were those pre green field benchmarks?]

After 10+ years of this, all of a sudden the CIO's of the world went ballistic when they found that 80-90% of their servers were running less than 10-20% utilization in peak times. And in some cases, the reason for the "high" 10-20% utilization was because it also caught the backup times in the overall calculations.

HP, IBM, Sun were all guilty of not recognizing that Customers were under such extreme pressures to reduce IT costs that they decided to move to commodity OS's/HW - NOT because the target platform was technically better, but rather deemed to be "good enough" and saved them some pretty big $'s. Course, they now have even bigger issues with VM sprawl, but that’s a different discussion.

That is what happened in the last 10 years.

In the next 10 years, the same thing is going to repeat itself - only with the software OS's and applications. Remember the old saying about Oracle - 80% of the Customers only use about 20% of all of the available features.

The Oracle, SAP's of the world are in for a big shock when there is a massive shift to SW solutions that cost much less, but are deemed to be "good enough".

Regards,

Kerry Main
Kerry dot main at starkgaming dot com