[RC5] Cartographie

Peter Cordes peter at llama.nslug.ns.ca
Thu Feb 10 23:02:30 EST 2000


On Thu, Feb 10, 2000 at 09:19:59AM -0000, Ben Ketteridge wrote:

> On the other hand, since we are talking about downloading 300KB+ we should be
> looking
> very seriously at compression technologies. 

 Good idea, that might help if the data has much of a pattern.

> But back to the original problem of serving 150kb/s from d.net. That's only the
> mean,
> I am sure (admin correct me if I'm wrong?) that clients do not keep up a
> consistent
> stream of contacts with the server, therefore, the serving rate must peak
> several times
> per day. This is probably quite predictably, perhaps (I am guessing here)
> around 9am,
> 5pm, 6pm and perhaps 11pm (local times around the world) - this may all even
> out of
> course, but I doubt it, due to the geographic distribution of clients.
> So the serving capacity will have to be much more than 150kb/s. Having said
> that, I
> used to get 120kb/s as a user on a shared university T1 link, so if d.net can
> get this kind
> of connection then the bandwidth may be achievable.

 That's what proxies are for.  Different proxies could fetch their daily
load of data at different times, staggered so that d.net's central server
only has to keep up a more or less constant 166kB/s.  (1E3 * 2^20 kB / 2
years.)  Of course, if d.net is to be useful, it should be _faster_ than what
they were projecting it would take.  In any case, proxies can average out
the bandwidth needs of the central server.  If there were several main
servers, they could each have a portion of the total data.

 Don't forget that the results comming back from the clients take bandwidth
too.  That about doubles the total throughput need we're predicting,
depending on how big the results are compared to the raw data.  All the
results have to come together in one place somehow, maybe by sneakernet with
_big_ tapes/whatever.  I think it makes sense to have more than one main
server, as well as having lots of proxies.

-- 
#define X(x,y) x##y
DUPS Secretary ; http://is2.dal.ca/~dups/
Peter Cordes ;  e-mail: X(peter at cordes.phys. , dal.ca)

"The gods confound the man who first found out how to distinguish the hours!
 Confound him, too, who in this place set up a sundial, to cut and hack
 my day so wretchedly into small pieces!" -- Plautus, 200 BCE

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list