You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@directmemory.apache.org by Tim Williams <wi...@gmail.com> on 2012/01/25 12:02:19 UTC

Performance testing infrastructure [was: Re: Release? [was: Re: Status of DirectMemory]

On Fri, Jan 20, 2012 at 10:31 AM, Raffaele P. Guidi
<ra...@gmail.com> wrote:
> Benchmarking with 4, 8 and 16 gb would be a good start. Existing tests
> could be used as a starting point but I think they will need changes and
> tuning - and I have personal problems at the moment that prevent me from
> being able to work on this.

I started a conversation over on infrastructure@.  The short version
is that we have no 'standard' way to support this right now but
they're willing to discuss it.  One option is to use EC2, but they
need more information on what's needed in terms of detailed ideas of
CPU cores, CPU time, disk space, disk i/O usage etc.  Obviously, they
can't just let us loose on EC2 without knowing what the bill would
look like in the end.  Can someone either give an idea of those
details (?maybe in terms of the EC2 instance types? and amount of time
needed)?  Or, you could join the conversation on infrastructure@...

Thanks,
--tim

Re: Performance testing infrastructure [was: Re: Release? [was: Re: Status of DirectMemory]

Posted by "Raffaele P. Guidi" <ra...@gmail.com>.
Sure, I used an extra large instance with 16gb of ram in the past - it
costed about 1 dollar per hour
Il giorno 25/gen/2012 12:02, "Tim Williams" <wi...@gmail.com> ha
scritto:

> On Fri, Jan 20, 2012 at 10:31 AM, Raffaele P. Guidi
> <ra...@gmail.com> wrote:
> > Benchmarking with 4, 8 and 16 gb would be a good start. Existing tests
> > could be used as a starting point but I think they will need changes and
> > tuning - and I have personal problems at the moment that prevent me from
> > being able to work on this.
>
> I started a conversation over on infrastructure@.  The short version
> is that we have no 'standard' way to support this right now but
> they're willing to discuss it.  One option is to use EC2, but they
> need more information on what's needed in terms of detailed ideas of
> CPU cores, CPU time, disk space, disk i/O usage etc.  Obviously, they
> can't just let us loose on EC2 without knowing what the bill would
> look like in the end.  Can someone either give an idea of those
> details (?maybe in terms of the EC2 instance types? and amount of time
> needed)?  Or, you could join the conversation on infrastructure@...
>
> Thanks,
> --tim
>