SBF queueing?

SBF queueing?

[Gregory Maxwell]

Has anyone decided to code a SFB (Stochastic Fair Blue) queue implementation
for Linux? It's been implemented for FreeBSD/ALTQ 
(http://www.eecs.umich.edu/~wuchang/blue/). The paper for it shows it
performing very well in comparison to RED.

It might be useful in a Linux implementation to be able to adjust what
fields are hashed (I don't believe the initial implementation does this,
though it has been a few months since I read the paper). 

For instance, on edge networks you hash src,dest,proto,sport,dport to
allocate fairly by flow. On larger networks, hashing on each flow will
require a fairly big filter to prevent collisions from punishing good flows
because of a collision with an unresponsive flow. It might be useful to only
hash on src on core networks, and potentially masked src on backbone transit
networks (which would have nice social implications as well, produce an
unresponsive flow and watch your entire subnet be slowed during networking
congestion, thus edge networks would implement technology to detect and
police unresponsive flows, something better done towards the edge). 

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: SBF queueing?

[patrick.mourlhon]

Hi Gregory!

You might have a look on linux/Documentation/networking/policy-routing.txt

I think this was down by Alexey Kuznetov

You might have a look to iproute + tc and HOWTO on advanced networking

patrick mourlhon

On Sat, 27 Jan 2001, Gregory Maxwell wrote:

> Has anyone decided to code a SFB (Stochastic Fair Blue) queue implementation
> for Linux? It's been implemented for FreeBSD/ALTQ 
> (http://www.eecs.umich.edu/~wuchang/blue/). The paper for it shows it
> performing very well in comparison to RED.
> 
> It might be useful in a Linux implementation to be able to adjust what
> fields are hashed (I don't believe the initial implementation does this,
> though it has been a few months since I read the paper). 
> 
> For instance, on edge networks you hash src,dest,proto,sport,dport to
> allocate fairly by flow. On larger networks, hashing on each flow will
> require a fairly big filter to prevent collisions from punishing good flows
> because of a collision with an unresponsive flow. It might be useful to only
> hash on src on core networks, and potentially masked src on backbone transit
> networks (which would have nice social implications as well, produce an
> unresponsive flow and watch your entire subnet be slowed during networking
> congestion, thus edge networks would implement technology to detect and
> police unresponsive flows, something better done towards the edge). 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> Please read the FAQ at http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: SBF queueing?

[Gregory Maxwell]

On Sat, Jan 27, 2001 at 07:52:32PM +0100, patrick.mourlhon@wanadoo.fr wrote:
> Hi Gregory!
> You might have a look on linux/Documentation/networking/policy-routing.txt
> I think this was down by Alexey Kuznetov

Thanks for the quick reply. But that's not exactly what I was looking for.
I was trying to find out if anyone was working on a peticular queueing
algorithim that Linux doesn't currently have.

> You might have a look to iproute + tc and HOWTO on advanced networking

You mean this "HOWTO on advanced networking"?
http://www.ds9a.nl/2.4Routing/

Look at the second authors name on the page. :)

> patrick mourlhon
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: SBF queueing?

[kuznet]

Hello!

> Has anyone decided to code a SFB (Stochastic Fair Blue) queue implementation
> for Linux?

I did not hear anything about this.

> (http://www.eecs.umich.edu/~wuchang/blue/). The paper for it shows it
> performing very well in comparison to RED.

Yes, the algorithm looks interesting.

Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/