- From: Simon E Spero <ses@tipper.oit.unc.edu>
- Date: Tue, 20 Dec 94 18:38:27 -0500
- To: Jeffrey Mogul <mogul@pa.dec.com>
- Cc: Henrik Frystyk Nielsen <frystyk@www0.cern.ch>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
After taking a longer look at the proxy problem it looks like there might be a work around based on using a different version number; I believe that proxy servers generate the header version themselves independent of the version given on the string. The end-server can be configured to ignore all connection headers unless the version is >1.0. The adaptive timesouts sound useful - there are very definite usage patterns that show up if log data is sorted into virtual sessions. Some patterns (like image fetches) are general, but there also some interesting local patterns ( if someone looks at the Dr Fun home page, they're very likely to get the cartoon of the day.) Do you have some more info about the data set you ran the traces on? I seem to recall you mentioning that the traces were taken from the California Election server over a couple of days? Do you have any information on the size of the data set, and the relative access frequencies of the various documents? Simon
Received on Tuesday, 20 December 1994 15:41:32 UTC