launchpad-dev team mailing list archive

[Robert Collins <robertc@xxxxxxxxxxxxxxxxx>]

On Tue, Mar 15, 2011 at 10:40 PM, Martin Pool <mbp@xxxxxxxxxxxxx> wrote:
> Robert asked the other day if the lp:kanban script I was running had
> got any faster because of lp's deployment changes to reduce queuing of
> http requests.
>
> It gone from 2:54 against old lpnet and 2:14 against old edge, down to
> 1:44.  (Run from my machine in Australia.)

-cool-

> However, the amount of work it does depends on the number of
> interesting bugs it processes, and that's changed since I last
> measured this in February in the context of
> <https://bugs.launchpad.net/bugs/712924>.
>
> So it's not an utterly reliable benchmark, but if it is valid being
> 40% faster is pretty nice!

reliability aside it sounds nice :)

> I still see quite a lot of variability in response time from one api
> request to the next.

If the the apis are the same method (perhaps on different bugs), then
much of the variation will be in-dc (or in-internet) latency. If they
are differnet apis - well, the performance of different api calls can
be wildly different.

Thanks for reporting back!

-Rob