[OSM-dev] scaling

[Frederik Ramm]

Hi,

Kai Krueger wrote:
> So unless you don't want to expand at all, thinking about scalability is an
> important issue and probably shouldn't be left till all of the servers are
> creaking and failing under the load.

Maybe you should re-read my post. I think there's a certain rate of 
growth that we can take, and I believe I wrote as much. What exactly 
that rate of growth is, is certainly something we can discuss, or we 
could also set some kind of sane target and then plan for that.

I don't think that I recommended to stop investing in hardware either.

> And it is not that we don't ever see any overload problems with our current
> setup even with current editing levels, or that everything is so massively
> over provisioned that one doesn't observe any slowdown during Sunday
> afternoon peak times.  

At last week's stratic working group IRC meeting, one of the items 
discussed was funding, and how money should be distributed to various 
working groups. My point in that discussion was that any money not 
desperately needed for something else in OSM certainly has to be 
allocated for server infrastructure.

> Frederik Ramm wrote:
>> So in a way, we must be glad that we have this technical limitation,
> 
> Making the API so slow that people don't want to edit anymore is not the
> right solution to the social issue, or to any other issue either.

You're dangerously close to trolling. The technical limitation I was 
talking about is what Steve said - that even with large-scale support we 
would not be able to grow 100fold in a matter of months. And I stand by 
that: I'm glad that we cannot do that technically, because that shields 
us from having to do it socially.

The German community which you cite as an example has grown to its 
current level slowly and steadily; and of course it is desirable, and 
manageable, for every other OSM community to do the same - but not 
within a week.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"