Have compression (was Re: Standards (was [BitTorrent] Back to

Olaf van der Spek OvdSpek at LIACS.NL
Wed Feb 9 06:45:00 EST 2005


Justin Cormack wrote:
> beginning (when most peers dont have anything) and the end (when most peers
> have the same as you), you can cut this latency by a huge amount by sending

Ah, you're assuming all peers are at the same stage. That's not what the 
normal end game mode in the official client is about.

> the have message for a smaller unit (like the minimum verifable unit), so
> that it can be sent along a chain of peers much faster).
> 
> This gives us another strategy for out have compression algorithm: if the peer
> is interested, dont send haves, batch them up so they can be coalesced into
> haves for larger amounts. After all the peer is already saying it is happy
> to download what we are offering. This reduces bandwidth used for have 
> messages even more.
> 
> Your solution of chunk-haves can be used for reducing latency, but doesnt
> suggest strategies for how to use it, and when you use it it always increases

True, but as different usage strategies don't (always) require different 
protocol syntax, I'd consider that less important.

> bandwidth. My version starts to suggest how to use it, and has other side
> effects like reducing the load on seeds and near-seeds as they dont have
> to send out bitmaps, they can send high order have messages.

Also true, but my extension was specifically designed to keep the 
changes as simple as possible.

Changes in the p2p protocol are (IMO) easier to make (via extension 
bits) then major incompatible changes in the .torrent format.


 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/BitTorrent/

<*> To unsubscribe from this group, send an email to:
    BitTorrent-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 





More information about the BitTorrent mailing list