On Aug 3, 2006, at 8:32 AM, Alex Tweedly wrote:
Yes, it does matter. There are a number of cases where multiple
connections will win out :
1. avoiding start/finish round-trip delays for each transfer.
You don't of course avoid them - but you can interleave them
with the parallel transfers, and hence keep the pipe closer to full.
2. transient congestion (and packet-drops) have less effect on
multiple connections
Esp. if the congested router is using RED
3. rate-limiting per connection from the server can be imposed
lower than your connection bandwidth
4. bandwidth allocation on congested links
Esp if the upstream router is using WFQ or DRR queuing, and
high capacity connections are severely limited
Thanks for these.
1. Interleaving is possible with TCP, but for the most part TCP does
not have round-trip delays except at the start and end. Well, I
guess some settings might. There will be some interleaving from what
I've seen.
2. I don't know much about congestion at routers, but I agree with
the affect of dropped packets.
3. I think rate-limiting at source, if you are downloading from
multiple servers, is the greatest factor.
4. I claim ignorance.
Another factor might be the ability to give user feedback right away
should there be a problem with any files.
I wonder if downloading multiple files might cause variations in
transfer rates that confuse the optimization at servers and routers.
Just mumbling.
Dar Scott
_______________________________________________
use-revolution mailing list
[email protected]
Please visit this url to subscribe, unsubscribe and manage your subscription
preferences:
http://lists.runrev.com/mailman/listinfo/use-revolution
