[Message Prev][Message
Next][Thread Prev][Thread Next][Message
Index][Thread Index]
Re: UKHA Team on Grid.org
- Subject: Re: UKHA Team on Grid.org
- From: "mark_harrison_uk2" <mph@xxxxxxxxxxxxxxx>
- Date: Sun, 26 Nov 2006 18:04:47 -0000
--- In ukha_d@xxxxxxx, yahoo@... wrote:
>
> Interestingly, I'm currently doing my masters on Grid computing too
> and I've been trying to think of some kind of fun HA project to
> run on my home Condor pool, or maybe something a bit more complex
> using Globus Toolkit and webservices. I kind of want to write
> something myself, not just participate in one of these desktop
> Grid projects, though I should join up to the cancer research while
> I think up ideas I suppose.
It would be an interesting (well, to someone like me) project to be
able to take a data feed from a wide range of [Harmony / Homeseer /
Homevision / xAP / xPL] users, that showed what they had done at
different times. (Obviously, there would need to be a bit of work
sorting out which log formats were suitable / readily transformable
into a common XML log - I suspect that xPL, xAP and Harmony would be
trivial.)
... then compare that to a set of external datafeeds [weather reports
/ TV listings / BBC news / etc.] and see whether there were any kind
of correlations that could be drawn...
... eg, people turn on their tellies when Stargate is about to start
on Sky, but turn them off their computers whenever Mark Harrison posts :-)
The reason that I think this is vaguely interesting from a
computational point of view, is that the "Grid mothership" would
only
need to aggregate the data feeds from the HA applications... but the
"external feeds" could be a series of RSS feeds from external
servers.
Hence the bandwidth requirements at the mothership would not be overly
large relative to the total size of the data being analysed.
What I'm not sure about, and hence, why this might be an interesting
Masters project from a "Grid-theory" point of view as well as an
HA
point of view, is whether the correlation modelling could be expressed
in a way that lent itself to distributed processing on a grid. (It may
well be that this problem has already been studied and there's
actually nothing novel there, though.)
Regards,
Mark Harrison, BA, MA, MBCS
[Declaration - the author has a financial interest in the Harmony
software]
PS - I don't normally use my letters, but given we were discussing
things academic :-)
** Sponsored by http://www.BERBLE.com **
all the Cool Stuff, in one Place
UKHA_D Main Index |
UKHA_D Thread Index |
UKHA_D Home |
Archives Home
|