- From: Tai Jin <tai@nexus.hpl.hp.com>
- Date: Tue, 3 Sep 1996 11:11:54 -0700
- To: peterson@austin.ibm.com
- Cc: fielding@avron.ICS.UCI.EDU, frystyk@w3.org, hallam@w3.org, jg@w3.org, mogul@pa.dec.com, timbl@hq.lcs.mit.edu, www-talk@w3.org
Another problem with callbacks is if the proxy is unreachable when the server performs the callback. And for scaling purposes, the server would have to restrict the use of callbacks to proxies (no browsers) or to some subset of proxies. The overhead of maintaining all the callback state is probably not worth the benefit of getting current content for most types of content. I would suggest a selective callback mechanism where only certain types of content use callbacks. Content which is updated on a regular schedule or have a known expiry date do not require callbacks since the Expires: header will work just fine for them. And most other content is not so important that serving a stale page is going to be a problem (e.g., my personal page). So callbacks should only be used for content which is updated in an adhoc manner and which cannot afford to be served stale without causing an inconvenience or worse (e.g., bus schedule information). The question is whether the amount of content in this category is small enough that a callback scheme could be feasible from a scalability perspective. And you still have to deal with the issues that were mentioned before. ...tai
Received on Tuesday, 3 September 1996 14:12:26 UTC