On behalf of the WMF, I wanted to share more information about what has happened and what our plan is moving forward. We were actually discussing how to communicate this when the thread popped up, and I apologize for not sending something out sooner.
Basically, the load on our maps servers increased to the point where the service was falling down. We had an incident in February where this happened and while the root cause is not well understood yet, it was clear that the issue is systemic and this was not a 1-time event. Details are
here. A big part of our map traffic (~35% is from 3rd parties) and by limiting support for third parties, we were able to mitigate the issue.
The immediate change we made was to drop cache misses from 3rd parties, but to keep serving the tiles that are in cache. This lessens our loads without turning 3rd party support off entirely, but is understandably confusing to the few 3rd parties who are using our service. In terms of long term support, we do not have any plans to increase it and are actively discussing how we might further limit 3rd party support in the future. We're doing this proactively so that we can provide appropriate warning, details and dates for any further decreases in service that might be coming down the pike.
As always your feedback is very welcome. What do you think? What are your primary questions, etc. We may not be able to respond right away as finding answers can sometimes require communication across many time zones, but that doesn't mean we're not listening and taking it in.
Thanks,
Jon
Director of Product Management, Wikimedia Foundation