Talk:Developer FAQ

From OpenStreetMap Wiki
Jump to navigation Jump to search

Discuss Developer FAQ


Is XAPI a good API substitute?

Frequently in response to people getting recommended or required to not use the main API due to heavy use, they get suggested to use XAPI instead. What makes XAPI any more suited for bulk usage than the main API? Indeed, it often appears to be overloaded itself. So is it more of an argument down the lines, well, if bulk users overload the XAPI, who cares, as it only affects other bulk users using XAPI and not editors? Or is there anything inherently more efficient about XAPI? If anything, I would have expected the recommendation to be towards using the TRAPI loadbalancer as thanks to the pre-tiled approach of TRAPI it can handle bbox requests much easier and with less hardware requirements than the db based APIs. Shouldn't XAPI resources be kept for people who actually need the extended capabilities of xapi rather than bogging it down with bbox map requests? --Amm 13:53, 14 September 2010 (BST)

My understanding was that the original XAPI somehow has/had the ability to return larger data areas without the bbox size restriction. How it scales to cope it such requests, or just more requests more often... well certainly the answer to that seems to be: it doesn't. I wondered at one stage, whether it would scale well because it streams the response so slowly. But no... it was just slow. These days we need to be more specific about which XAPI we're talking about of course. Pretty sure the new jXAPI instances won't cope with tonnes of traffic particularly well either though. Tiled data could be the way forward for this kind of thing I would've thought. - Harry Wood 01:04, 3 February 2012 (UTC)