[Osmf-talk] API up after maint. was: Two day of down server is not possible !!!!

Andy Robinson (blackadder) blackadderajr at googlemail.com
Mon Aug 24 08:29:18 UTC 2009


Frederik Ramm wrote:
>Sent: 23 August 2009 9:12 PM
>To: Andrew Turner
>Cc: osmf-talk at openstreetmap.org
>Subject: Re: [Osmf-talk] API up after maint. was: Two day of down server is
>not possible !!!!
>
>Hi,
>
>Andrew Turner wrote:
>> I would like to discuss how we could address downtimes in the future.
>> I understand that there is a "primary objective" of OSM.org being used
>> for editing & maintaining data. However, as OSM becomes more
>> "mainstream" there is definitely an expectation that it will be up and
>> reachable at any time by the general public.
>
>We could work on educating the public so that the expectation is
>removed, rather than do what the mainstream expects from us.
>
>I'm not totally against what you're suggesting but I sense some
>"slippery slope" here; maybe it is time to get the message across more
>clearly that we are not in anyway a Google Maps replacement and if
>people expect this from us they should change their expectations.
>(Because they'll be "expecting" routing, and aerial imagery, and "MyMap"
>features, and all that in no time if we don't set the record straight.)
>
>I'm sure that with some effort, this could be said in a friendly way
>that is actually understood by the casual visitor.

I agree. The Foundation's current aims (subject to any change in focus made
by the membership and the wider OSM community) is not to be a map tile
delivery service. Our platform, hosting and services, are not configured in
such a way that tile serving is our primary focus, it's defiantly a
secondary focus.

It should also be understood that delivery of tiles doesn't come cheap if
folks want a guaranteed service. UCL, our primary host, donates the
bandwidth we use for the services hosted at their facility, including
currently the tile server. We have no formal SLA with UCL and we therefore
have to live with the occasional circumstances when UCL has to shut down the
network. Currently paying for bandwidth and hosting of tiles at a guaranteed
level is beyond the funding means of the Foundation and therefore till now
we have made it clear policy that tile delivery to end users should be
managed by others. Our own tile delivery is designed to provide a shop
window to the project, not a wide scale service to third party applications.

I'm not sure whether it's been stated clearly in this thread but it was UCL
that effectively shut down the servers this weekend. They scheduled some
electrical work which necessitated our platform shutdown. Our sysadmins then
took the logical decision to use the opportunity to undertake some hardware
maintenance of their own.

As Frederik says, it's clear we need to better educate the wider userbase
what the limitations of the OSM platform are and probably provide better
guidance to those folk who are developing services off the OSM platform or
data. The expectations of some are clearly higher than that which is
achievable from our presently configured platform.

Cheers

Andy





More information about the osmf-talk mailing list