[Info-vax] OT: London stock exchange switches to Linux

[John Wallace]

On Jan 28, 11:47 am, Neil Rieck <n.ri... at sympatico.ca> wrote:
> On Jan 25, 2:52 pm, John Wallace <johnwalla... at yahoo.co.uk> wrote:
>
> [...snip...]
>
> > Windows may not have been up to the job, but a "soft real time" OS
> > such as a Linux with a pre-emptible kernel is probably entirely
> > capable of "good enough" performance. VMS could do it too, but that's
> > another story, one seemingly of insufficient interest to HP and its
> > "partners".
>
> > There's nothing particularly new in this LSE story, some folks have
> > been doing this kind of thing for a year or three. Like HP, for
> > example, and Novell/SuSe, in this 2008 press release (there were other
> > similar ones from other x86 folk before and after this one). The HP
> > group in this picture is of course the Proliant folks, with x86-64
> > blades. Note that this press release isn't a trading application, it's
> > a Reuters messaging thing, but there are similarities in the
> > requirements:
>
> [...snip...]
>
> I was partially-flamed for the initial post so let me add one or two
> thoughts.
>
> I don't think the LSE went out looking for a Linux (or, previously,  a
> Windows 2003) based system. Some contractor-based company approached
> them with a business plan based upon this OS. Now I am as big a fan of
> OpenVMS as anyone else around this NG but let's face it, for some
> reason companies don't seem to be offering new solutions based upon
> OpenVMS. Is it because of the the price of the Itanium hardware
> (compared to x86-64) or is it the price of software licensing? Who
> knows? But I think everyone would agree that HP is not marketing their
> OSs in the same way that companies did 25 years ago.
>
> On a different note, I read an article last year indicating that
> Google is the forth largest PC manufacturer in the world. Google? Yes,
> they make machines to their own spec for their own consumption. Now
> everyone knows that "Google searches" are ridiculously fast and that
> their system is based upon a large array of PCs connected in a cloud/
> grid configuration. So it makes you wonder if someone ever tried a
> similar approach for doing massive amounts of limited function
> transactions (like buy/sell stocks) using a cheap/free OS.
>
> NSR

Google (search) don't do what folks round here would call
"transactions". Mostly Google searches do read only lookups where only
Google know what the "right" answer is and where no one really cares
if a partial (rather than complete) answer is returned, or even if an
incorrect answer is returned. Occasionally you get total silliness
like "Showing results 1-3 of 0".

Now try that approach with (say) medical records, payroll data, or
buying/selling stocks. It doesn't translate. That does not of itself
disqualify the grid approach for doing massive amounts of limited
function transactions, but the classic problem with that approach is
that there is usually a serialisation/synchronisation point somewhere
if it's a genuine "transaction, and as soon as you have that to
address, massively distributed systems become pointless because the
overhead of synchronisation exceeds the benefit from the extra
systems. On the other hand a small number of properly designed
distributed systems can bring some added value in terms of performance
and resilience.

On the other hand a grid/cloud approach is potentially perfectly valid
for a workload which consists of sets of (almost completely)
independent tasks to be done (e.g. like SETI at Home, or more recently
Google Apps over the cloud).