What is an acceptable use of the Confluence xml-rpc API?

[Mick Watson]

Every hour we run a Python script that produces approx. 800 API queries.  These download the contents of a page and do some analysis on them.

Our provider is telling us that this is excessive use and that our queries are contributing to a degradation of the service.

Can anyone comment on what they consider an acceptable use?

What are others using the API for?

[Jonathan Simonoff]

AFAIK reading a page via the API is pretty much equivalent to accessing the page via a browser, as far as load on the server.  So, if you're reading 800 pages an hour, that is like users reading 800 pages.

Is that a lot of use for your installation?  If so, it may excessive.

[Mick Watson]

Absolutely not. Confluence is supplid as a service to our entire University, with literally 1000s of spaces and 1000s of users.

[Jonathan Simonoff]

It is hard for me to tell whether or not it is really a big load on your server. If you are doing it 24x7, of course, that would be almost 20,000 page reads a day from this script. If you have 10,000 users but they only read a couple of pages a day each on average, then you are doubling the load. An additional factor might be other operations you are doing aside form reading pages. Some operations (like listing all pages in a large space) can be fairly expensive. Anyway, in the grand scheme of things adding 800 page reads an hour to a large installation isn't an awful lot, but I can see why your provider might see that as excessive use from one account, and be concerned about it. Were I your admin, I would be concerned about that kind of use from an ordinary user -- if this is some sort of special, unique operation, it might be OK, but if there are other people wanting to do similar things, that seems like more of a problem. Anyway, there is no real definition of "excessive use". It depends on what else is going on, the configuration of the installation, your network, etc.