- From: Jim Seidman <jim@spyglass.com>
- Date: Fri, 16 Dec 94 09:58:30 -0600
- To: John Franks <john@math.nwu.edu>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
John Franks writes: >Correct me if I am wrong, but I concluded from Spero's postings that >nothing currently proposed including MGET, hold-open, or even HTTP-NG >would improve (or even match?) the user's perceived performance >currently given by Netscape. By this I mean the ellapsed time until >the user can start reading *all the text* and the ellapsed time until >the user can jump to a new link. I have to disagree here. If the issue is how quickly all the text can be downloaded, then the fastest way to do this is to send only the text and then pass on the images later. Especially for users on slower connections (such as the ever-mentioned 14.4 SLIP connection) there is a notable degradation in getting all of the text if several images are being sent simultaneously. There are several other reasons why multiple simultaneous connections a bad idea. As you pointed out, there are many aborted connections. When downloading one item at a time you are much more likely to have received complete items which you can cache. (For example, if you are downloading a page with 15 images which are reused on other pages throughout the site, you might at least get a few of them completely, rather than getting several useless partial images which you'll then reload from scratch when they appear on the next page.) In my mind the only defensible reason for simultaneous connections is to reduce the round trip time penalty for loading all the pieces of a document. However there are already several discussions taking place here about other ways to minimize the RTT issues. Even something as trivial as having a keep-alive and allowing the client to buffer up requests for images as it's downloading the text would provide a major improvement. -- Jim Seidman Senior Software Engineer Spyglass, Inc.
Received on Friday, 16 December 1994 07:59:52 UTC