- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Wed, 07 Aug 96 16:00:22 MDT
- To: Koen Holtman <koen@win.tue.nl>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
But to allocate max-uses values to proxies an efficient way, an origin server seems to have to keeping per-proxy database of `max-use-qouta-use-speed' (last two paragraphs of Section 2), which adds some overhead to every request. Only if it really wants to be cautious about bounding the counting error. I would imagine that the simplest servers would have a global setting for this value (e.g., always send "max-uses=10"). A somewhat more sophisticated server might keep a moderate-sized cache of (proxy-address, max-use-setting) entries. At about 8 bytes per entry (4 bytes IP address, 2 bytes max-use-setting, perhaps 2 bytes for an LRU counter), you could spend about a dime on RAM and keep 1000 such entries. Reading these paragraphs, the goal of the max-uses allocation heuristics seem to be to ensure that all counts are reported `soon enough'. "Soon" in the sense of "bounding the error in the final count", not in the sense of "within 3 hours". It seems that a max-time-to-wait-before-reporting-hits mechanism, can achieve the same goal without the same computational overhead in origin servers. This mechanism would also eliminate the need for implementing difficult max-use distribution heuristics in proxy caches: a cache could simply subtract the age of the response from the max-time value. Even better, we *already have* a max-time-to-wait-before-reporting-hits mechanism in the form of cache-control: max-age. I conclude that the max-use mechanism is unnecessary and propose that it is removed, and that a section about using cache-control: max-age is added. I think this would be a mistake, since it doesn't distinguish between very busy caches and very lightly-used ones. If you send a small max-age value, then the lightly-used caches might report far more often than necessary. If you send a large max-age value, then the heavily-used caches might report far too infrequently (in terms of number of uses between reports). -Jeff
Received on Wednesday, 7 August 1996 16:11:58 UTC