[LARTC] scheduler testing

[LARTC] scheduler testing

[Amit Kucheria]

Hi all,

I am currently trying to test the schedulers available in Linux - CBQ, 
FIFO, PRIO, SFQ and WRR.

My setup is simple -

 Source --------- Router --------- Sink
               eth0   eth1

- The links are 10 Mbit links.
- I am generating average traffic of 1Mbps at the source
- I have to throttle/rate-limit the router-sink link to 1Mbps to make it 
  look similar to a T1 link to the internet.
- I need a rate-limiting architecture which will allow be to test various 
  schedulers as well as impose a strict limit on the outgoing link.
- I hope to cause queueing due to the statistical nature of traffic 
  (TCP,UDP) so that instantaneous source thruput might be greater than 
  1Mbit/s.

I thought of using a TBF class inside a CBQ root qdisc to rate-limit 
traffic, but that would drop extra traffic which would prevent queuing.

I would like ideas on how to create queuing assuming that the outgoing 
link is like a T1 link. Also what kinds of traffic can i use to simulate 
real life situations?

Regards,
Amit
-- 
The statement below is true.
The statement above is false.
^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^
                  Amit Kucheria
          EECS Grad. Research Assistant
         University of Kansas @ Lawrence
   (R)+1-(785)-830 8521 ||| (O)+1-(785)-864 7774
____________________________________________________


_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://ds9a.nl/2.4Routing/

Re: [LARTC] scheduler testing

[bert hubert]

On Fri, Dec 07, 2001 at 03:46:19AM -0600, Amit Kucheria wrote:
> Hi all,
> 
> I am currently trying to test the schedulers available in Linux - CBQ, 
> FIFO, PRIO, SFQ and WRR.
> 
> My setup is simple -
> 
>  Source --------- Router --------- Sink
>                eth0   eth1
> 
> - The links are 10 Mbit links.

Look at www.docum.org, Stef did a lot of cool work in testing.

> I thought of using a TBF class inside a CBQ root qdisc to rate-limit 
> traffic, but that would drop extra traffic which would prevent queuing.

No, it will first queue quite a lot :-)

> I would like ideas on how to create queuing assuming that the outgoing 
> link is like a T1 link. Also what kinds of traffic can i use to simulate 
> real life situations?

The HTB site has some stuff: http://luxik.cdi.cz/~devik/qos/htb/

Regards,

bert

-- 
http://www.PowerDNS.com          Versatile DNS Software & Services
Trilab                                 The Technology People
Netherlabs BV / Rent-a-Nerd.nl           - Nerd Available -
'SYN! .. SYN|ACK! .. ACK!' - the mating call of the internet

_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://ds9a.nl/2.4Routing/

Re: [LARTC] scheduler testing

[Amit Kucheria]

On Fri, 7 Dec 2001, Amit Kucheria wrote:

Hi all,
 
I am currently trying to test the schedulers available in Linux - CBQ, 
FIFO, PRIO, SFQ and WRR.
 
My setup is simple -

          10Mbit           10Mbit
  Source --------- Router --------- Sink
                eth0   eth1
 

- I need to simulate the 'router-sink' link at a T1 link. obviously the 
best way to do this is to but a TBF on eth1.

But i want to test various kinds of queueing disciplines with this rate 
limited traffic. This wont be allowed by TBF which cant take classes.

So would the following work ?

                    1Mbit
TBF qdisc 2:0     +--------+    *******     My various classes FIFO, CBQ
CBQ class 1:1    +----------+  +--------+   CBQ class 1:2
                +-------------------------+  CBQ qdisc 1:0


Now TBF rate limits to 1Mbit and then a filter filters the packet to "my 
various classes which could be multiple CBQ, prio classes or simple fifo.

 - I am generating average traffic of 1Mbps at the source
 - I need a rate-limiting architecture which will allow be to test various 
   schedulers as well as impose a strict limit on the outgoing link.

Regards,
Amit

-- 
The statement below is true.
The statement above is false.
^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^-^
                  Amit Kucheria
          EECS Grad. Research Assistant
         University of Kansas @ Lawrence
   (R)+1-(785)-830 8521 ||| (O)+1-(785)-864 7774
____________________________________________________


_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://ds9a.nl/lartc/