Thank you James for your answer! > > AFAIK range requests are working reasonably well in the 3.2 release. The > next step that we would like is to be able to serve range requests out of > the cache while the object is being cached; but for many (most?) cases the > current support should be fine. You ought to be able to determine whether > the current support is sufficient for your needs by running a few tests. > Please file bugs if you encounter any problems. >
Sorry I wasn't clear. My goal is to understand first *how* currently ATS does range request caching: - does it cache partial objects (the parts corresponding to the requested ranges) or the whole requested file? - if it may cache partial objects is it be able to serve a range by taking subrange data from the cache (for already cached subrange) and other part from the requested file? - what is the storage used for caching range requests (does it share the same storage with regular objects or it has some special storage) - any other important aspects one might think important regarding range requests caching ATS implementation. > There's a bunch of information about the cache in the documentation, < > http://trafficserver.apache.org/docs/trunk/admin>. > The only reference about range requests there was in the "records.config" and it was about how to enable the range requests lookup in the cache: *>> proxy.config.http.cache.range.lookup* Best regards, Bogdan Graur
