Re: Yes ,you are right!

From: Alex Rousskov <rousskov@dont-contact.us>
Date: Mon, 30 Aug 1999 21:19:14 -0600

On Tue, 31 Aug 1999, Dancer wrote:

> big_tiger@263.net wrote:
> >
> > Yes,Netants do use the method you refered to.For example,on downloading a file,
> > 1st ant start at 0 byte,and 2nd ant start at 0+x byte,and 3rd at 0+x+x byte,etc.
> > Maximum 5 ants can be use simutaneously in one downloading file.Any idea to
> > solve my problem?Thank you!
>
> Start them in reverse order. So that you're starting the last range
> first. That will stop your problem. The resulting object won't be stored
> in the cache, though, I shouldn't think.

If each "ant" is requesting a single "simple" [x, y] range, then the order
is probably not significant for caching, and default order may give you
best response time. You may need to start your "ants" with some
interarrival delay though. Squid may be able to reply with cached partial
content then. However, there may be some heuristics added to the code that
would prevent even that. Your best bet is to (a) make sure that content is
cachable and range requests are "simple" and (b) experiment, I guess.

Alex.
Received on Mon Aug 30 1999 - 21:33:53 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:48:08 MST