About pluggable congestion control infrastructure in DCCP

About pluggable congestion control infrastructure in DCCP

[Shahiduzzaman]

Hi all,

This may be a redundant or very naive question - sorry in advance for
that. Can anybody tell me, whether the current Linux kernel DCCP
implementation is done in a fashion like TCP pluggable congestion
avoidance modules infrastructure (http://lwn.net/Articles/128681/) ?


Thanks.


Shahid

Re: About pluggable congestion control infrastructure in DCCP

[Ian McDonald]

On 8/20/07, Shahiduzzaman <shahid21st@gmail.com> wrote:
> Hi all,
>
> This may be a redundant or very naive question - sorry in advance for
> that. Can anybody tell me, whether the current Linux kernel DCCP
> implementation is done in a fashion like TCP pluggable congestion
> avoidance modules infrastructure (http://lwn.net/Articles/128681/) ?
>
>
> Thanks.
>
>
> Shahid

The short answer is yes and no!

No that the in-tree version does not support it.

Yes in that for CCID2 (TCP like) there was an out of tree version that
could use the different TCP congestion control methods.

If you mean a simple way to plug in new CCIDs like CCID4 etc then the
answer is no. However two people are working on CCID4 which may make
the code more modular.

Hope this helps,

Ian
-- 
Web1: http://wand.net.nz/~iam4/
Web2: http://www.jandi.co.nz
Blog: http://iansblog.jandi.co.nz

Re: About pluggable congestion control infrastructure in DCCP

[Arnaldo Carvalho de Melo]

Em Mon, Aug 20, 2007 at 08:53:15AM +1200, Ian McDonald escreveu:
> On 8/20/07, Shahiduzzaman <shahid21st@gmail.com> wrote:
> > Hi all,
> >
> > This may be a redundant or very naive question - sorry in advance for
> > that. Can anybody tell me, whether the current Linux kernel DCCP
> > implementation is done in a fashion like TCP pluggable congestion
> > avoidance modules infrastructure (http://lwn.net/Articles/128681/) ?
> 
> The short answer is yes and no!

I think the answer is just "yes"

The CCID infrastructure in the Linux kernel is done in a fashion like
the TCP pluggable congestion avoidances modules infrastructure. One can
write a new CCID and load, making it available for userspace programs to
use without changing a line of code in what is already in the kernel.
 
> No that the in-tree version does not support it.
> 
> Yes in that for CCID2 (TCP like) there was an out of tree version that
> could use the different TCP congestion control methods.

The question was if it was done "in a fashion like", not "exactly as". I
may be confused with your answer as english is not my first language tho
:-)

> If you mean a simple way to plug in new CCIDs like CCID4 etc then the
> answer is no. However two people are working on CCID4 which may make
> the code more modular.

Why not? See the recent implementation of CCID0 (UDP Like Congestion
Control :-)) by Gerrit:

http://www.mail-archive.com/dccp@vger.kernel.org/msg02091.html

Of course, as in the current TCP pluggable infrastructure it may well be
the case that the current CCID pluggable infrastructure something that
a new CCID wants to plug into is not pluggable yet.

- Arnaldo

Re: About pluggable congestion control infrastructure in DCCP

[Ian McDonald]

On 8/20/07, Arnaldo Carvalho de Melo <acme@ghostprotocols.net> wrote:
> Em Mon, Aug 20, 2007 at 08:53:15AM +1200, Ian McDonald escreveu:
> > On 8/20/07, Shahiduzzaman <shahid21st@gmail.com> wrote:
> > > Hi all,
> > >
> > > This may be a redundant or very naive question - sorry in advance for
> > > that. Can anybody tell me, whether the current Linux kernel DCCP
> > > implementation is done in a fashion like TCP pluggable congestion
> > > avoidance modules infrastructure (http://lwn.net/Articles/128681/) ?
> >
> > The short answer is yes and no!
>
> I think the answer is just "yes"
>
> The CCID infrastructure in the Linux kernel is done in a fashion like
> the TCP pluggable congestion avoidances modules infrastructure. One can
> write a new CCID and load, making it available for userspace programs to
> use without changing a line of code in what is already in the kernel.
>
> > No that the in-tree version does not support it.
>
> The question was if it was done "in a fashion like", not "exactly as". I
> may be confused with your answer as english is not my first language tho
> :-)
>

I think my answer may have been confusing then...

To clarify the CCID code is modular in that you don't need to write
the code from the ground up for a new CCID. It has callback functions
for implementing various protocol requirements.

The part where it is not modular (in the tree anywhere) is if you want
to change just the very small part of how the congestion control is
done for any particular CCID.

I hope this clarifies a little. I get confused even though English is
my first language!

Ian
-- 
Web1: http://wand.net.nz/~iam4/
Web2: http://www.jandi.co.nz
Blog: http://iansblog.jandi.co.nz

Re: About pluggable congestion control infrastructure in DCCP

[Arnaldo Carvalho de Melo]

Em Mon, Aug 20, 2007 at 10:22:40AM +1200, Ian McDonald escreveu:
> On 8/20/07, Arnaldo Carvalho de Melo <acme@ghostprotocols.net> wrote:
> > Em Mon, Aug 20, 2007 at 08:53:15AM +1200, Ian McDonald escreveu:
> > > On 8/20/07, Shahiduzzaman <shahid21st@gmail.com> wrote:
> > > > Hi all,
> > > >
> > > > This may be a redundant or very naive question - sorry in advance for
> > > > that. Can anybody tell me, whether the current Linux kernel DCCP
> > > > implementation is done in a fashion like TCP pluggable congestion
> > > > avoidance modules infrastructure (http://lwn.net/Articles/128681/) ?
> > >
> > > The short answer is yes and no!
> >
> > I think the answer is just "yes"
> >
> > The CCID infrastructure in the Linux kernel is done in a fashion like
> > the TCP pluggable congestion avoidances modules infrastructure. One can
> > write a new CCID and load, making it available for userspace programs to
> > use without changing a line of code in what is already in the kernel.
> >
> > > No that the in-tree version does not support it.
> >
> > The question was if it was done "in a fashion like", not "exactly as". I
> > may be confused with your answer as english is not my first language tho
> > :-)
> >
> 
> I think my answer may have been confusing then...
> 
> To clarify the CCID code is modular in that you don't need to write
> the code from the ground up for a new CCID. It has callback functions
> for implementing various protocol requirements.
> 
> The part where it is not modular (in the tree anywhere) is if you want
> to change just the very small part of how the congestion control is
> done for any particular CCID.

That needs improvements, yes, but even in this area we have separated a
good number of routines in the tfrc library:

[acme@mica net-2.6.24]$ l net/dccp/ccids/lib/
total 64
drwxr-xr-x 2 acme acme  4096 Aug 19 19:25 ./
drwxr-xr-x 3 acme acme  4096 Aug 19 19:26 ../
-rw-r--r-- 1 acme acme  7391 Aug 19 19:25 loss_interval.c
-rw-r--r-- 1 acme acme  1019 Aug 19 19:25 loss_interval.h
-rw-r--r-- 1 acme acme   119 Aug 19 19:25 Makefile
-rw-r--r-- 1 acme acme  7806 Aug 19 19:25 packet_history.c
-rw-r--r-- 1 acme acme  5853 Aug 19 19:25 packet_history.h
-rw-r--r-- 1 acme acme 18708 Aug 19 19:25 tfrc_equation.c
-rw-r--r-- 1 acme acme  1171 Aug 19 19:25 tfrc.h
[acme@mica net-2.6.24]$
 
> I hope this clarifies a little. I get confused even though English is
> my first language!

Yes, it clarifies, and the discussion should provide the information
that Shahiduzzaman is looking for.

Ok, back to preparing a batch of patches for upstream submission, in my
glacial pace, but hey, global warming will eventually reach my feets!
:-)

- Arnaldo