Re: Throttling cached content.

On jue, 2007-11-22 at 14:35 +1300, Amos Jeffries wrote:
> > On Wed, Nov 21, 2007, German Gomez wrote:
> >> Hello,
> >>
> >> I've reading through the list and found several references to this
> >> subject, how to limit bandwidth for cached content, and the only
> >> solution
> >> was to use QOS inside linux kernel. We have a remote accelerator proxy
> >> and would like to limit the bandwith for big files so smaller ones get
> >> dispatched as fast as possilbe but large one do not fill up your
> >> bandwidth.
> >
> > ok.
> >
> >> I have been reading through the code of delay_pools.c and it seems that
> >> it
> >> should be possible the use the same infrastructure for limitting the
> >> server bandwidth shouldn't it??
>
> The code itself maybe. However the side-effects of using it should be
> checked and tested well.
> I believe the data is buffered by the kernel on receipt to a large degree
> these days, whether the client app reads it out or not. Throttling the
> inbound side of the transaction would cause these buffers to fill faster
> than emptied and an overflow can be expected at some point outside squid.
> Whether any specific OS handles that nicely or not is an ongoing
> networking problem.
> The catch-22 is that throttling in this manner in a client app does not
> actually save any bandwidth until the OS buffer is has reached overflow.
>
> My opinion is that the joint application of collapsed-forwarding and
> improved caching is the better approach to take when server-side bandwidth
> is a consideration.

I don't know if I understand correctly your answer or if you have
understand our needs, we don't need to accelerate small files we need to
delay packet delivering for large files. We want to control outgoing
bandwidth from squid to final clients. Kernel buffers (at squid server)
should get filled slower this way, the only problem we could find would
be too many connections, but we should be able to control this balancing
load among several servers. About collapsed-forwarding it seems to
control bandwidth at the client side of squid, I mean 'it should limit
apache bandwidth usage' but not cached content delivery bandwidth.

If not ongoing project is working on it I think we should start looking
through the code to evaluate how hard would be to implement it. Also
would like to know who would be the ideal person for funding if we have
no time/knowledge to do it.

Regards,

- german

-- 
German Jose Gomez Garcia  | Mundinteractivos - El Mundo      |
Area de Internet          | Pradillo, 42                     |
german@el-mundo.net       | 28002 - Madrid (SPAIN, EU)       |
http://www.elmundo.es/    | Tel: (+34) 915864800 (Ext: 4616) |