Search the Archives

RFC 2068: I have realized that point 9.1.2 of the RFC 2068 is a big...

<< Back to: RFC 2068

Reader comments:

Comment by Doolyo
Submitted on 8/13/2005
Related RFC: RFC 2068
Rating: Not yet rated	Rate this comment:
I have realized that point 9.1.2 of the RFC 2068 is a big threat to the speed of the display of HTML pages on the internet.This restriction of maximum 2 simultaneous connections per server at the client level is highly visible on sites where we have a slow response time.I give an example:I have a server which has 40ms of response time, because in the same country than I am. The total load time of the home page is 2.3 seconds.Now if I have the same site in another country where the response time is about 200ms, the total load of the page is about 5.8 seconds!!This is about 3 times slower! This is not due to the bandwidth which is far high enough. I have analysed the traffic and these 200ms are always waited for, before sending another request from the client, and thus it is just normal to be so slow.Of course we expect the page to load slower because of this response time slowness. However so much is not normal because in theory, with the pipeline connections described in this document, we could go down up to just having 160ms more than for the other faster server (with infinite simultaneous connections). This would make 2.46 sec instead of 2.3 sec and there would be no difference at all for the user. Now, because of this restriction users must see their pages displayed very slowly when accessing to sites located far away from where they live.To me this is a big regression to the HTTP protocol. With HTTP 1.0 we had 4 simultaneous connections suggested, but now it goes down to 2 connections with HTTP 1.1. This makes HTTP 1.1 be much slower than HTTP1.0. Too sad...I also have tried to change in the registry for internet explorer in the HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Internet Settings\MaxConnectionsPerServer key and putting there 4 instead of 2. The difference is huge ! I have about 3.3 seconds instaed of 5.8, which is about the half of the load time !I remind that the server in question can be as fast as he could be, with the best bandwidth quality you could have, the result we still be the same: the page loads wil be about 2-3 times slower when connecting from abroad with 2 sim connections.Therefore, my strong suggestion is to increase this value from 2 to 4 like it was already the case vefore in HTTP 1.0.I hope my message will be heared. Else give a try and see for yourself. The congestion should not suffer by passing from 2 to 4. It could if we would put 10 here, but not 4. I think that this change is really necessary for the sake of the internet's reliability and easiness to use.Regards,Daniel

FAQS.ORG makes no guarantees as to the accuracy of the posts. Each post is the personal opinion of the poster. These posts are not intended to substitute for medical, tax, legal, investment, accounting, or other professional advice. FAQS.ORG does not endorse any opinion or any product or service mentioned mentioned in these posts.

<< Back to: RFC 2068