[Gllug] Lots of computing power required - is Amazon EC2 an answer?

[Sanatan Rai]

On 30 June 2011 23:52, John Hearns <hearnsj at googlemail.com> wrote:
> On 30 June 2011 23:42, John Hearns <hearnsj at googlemail.com> wrote:
>>
>>
>> At a first glance, depending on how big your XML dataset is of course,
>
> Sorry - 800 megabytes. Shouldn't take too long to transfer to
> workstations in a Condor pool.

EC2 will not give you a cluster to parallelise your programme over by default.
All you get is a machine to which you have remove access, and the best spec for
a single machine might not necessarily be better than what you own or
already have access to.

If you want to parallelise computations, then you will have to do the legwork
to set up enough EC2 instances. You might get lucky and there might be some
images already available, but you will still have to pay for all the instances.
So if all you need are a lot of machines for a short period of time, then indeed
EC2 is the way to go. But you will still have to set everything up,
and along with
data transfer etc the cost will not be trivial, though a lot cheaper than buying
new machines.

Are you absolutely certain that you cannot see an `algorithmic way out of this.'

If it doesn't violate your project's ethics policy, then perhaps you
could describe
the problem you're trying to solve, and then perhaps someone might be able
to suggest an alternative.

--Sanatan

-- 
Sanatan Rai
--
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug