Re: HTTP 1.1 pipelining & persistent connections query

From: Venkat Padmanabhan ([email protected])
Date: Thu Jul 31 1997 - 19:11:31 EDT


Matt,

  What you say is true: in some cases, it may well be faster to
launch several parallel connections than use P-HTTP (pipelining
over a persistent connection). For one thing when you launch 4
parallel TCP connections you are in effect starting off with
an initial window size four times as large as a single connection
would, so it is not suprising that you do better. But the real
issue is that if a lot of people started using the network in this
more aggressive fashion, they might just end up hurting themselves
and everyone else by pushing the network into a congested state.
One reason for using a single persistent connection is precisely to
avoid such a problem.

-Venkat

> Form: Memo
> Text: (48 lines follow)
> Can someone explain pipelining and persistent connection benefits of HTTP
> 1.1 to me?
> I have set up a simple spreadsheet to analyse this and get contradictory
> results.
> I have one column which is a single persistent TCP connection to transfer
> 1MByte, it takes 20 RTTs as can be seen in second column.
> I then have four parallel TCPs each transferring 130.5 kBytes taking 8 RTTs.
> I then consider 1 RTT for FIN and 3 RTTs for 3way handshake start. Four
> more TCPs then start and transfer 130.5 kBytes each in 8 RTTs. The total
> for 8 TCPs is 1.04448 MBytes and the single TCP total is 1.048064 MBytes.
> However the 8 TCPs finish before the single TCP. How then is it preferable
> to pipeline a single TCP rather than transmit multiple parallel TCPs such as
> HTTP 1.0 does?
>



This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:28 EST