Re: [aqm] AQM schemes: Queue length vs. delay based

[Preethi Natarajan <[e-mail]>]

From:  Naeem Khademi <[e-mail]>
Date:  Thursday, November 14, 2013 1:06 PM
To:  Preethi Natarajan <[e-mail]>
Cc:  curtis <[e-mail]>, Michael Welzl <[e-mail]>,
"[e-mail]" <[e-mail]>, Anoop Ghanwani <[e-mail]>
Subject:  Re: [aqm] AQM schemes: Queue length vs. delay based

> 
> This should have probably been brought in different thread...
> 
> comments follow inline
> 
> On Thu, Nov 14, 2013 at 9:00 PM, Preethi Natarajan <[e-mail]>
> wrote:
>> 
>> 
>> From:  Naeem Khademi <[e-mail]>
>> Date:  Thursday, November 14, 2013 8:05 AM
>> To:  Anoop Ghanwani <[e-mail]>
>> Cc:  <[e-mail]>, Michael Welzl <[e-mail]>,
>> "[e-mail]" <[e-mail]>, Preethi Natarajan <[e-mail]>
>> 
>> Subject:  Re: [aqm] AQM schemes: Queue length vs. delay based
>> 
>>> 
>>> 
>>> Below is my personal opinion, but hopefully Fred can clarify this better
>>> based on the AQM recommendations draft:
>>> 
>>> "applicability of AQM to all types of networks/switches" => "yes"
>>> 
>>> "applicability of *any* AQM to all types of networks/switches" => "no"
>>>  
>>>> Perhaps AQM cannot help this, but hopefully it won't hurt.  Trying to do
>>>> fancy things with small buffers is challenging.
>>> 
>>> AQM will most likely to help from data centers to the access links, and so
>>> on. But we may possibly need different AQMs for different network scenarios;
>>> The fact that an AQMs should be auto-tunable doesn't imply that it can be
>>> applied everywhere and we may need different auto-tunable AQMs specifically
>>> designed for different networks (ideally better if we could use fewer of
>>> them and they could work everywhere). I hope I'm not wrong.
>>>  
>> 
>> Again, please hold on. Even if its personal opinion, I would state facts
>> backed with evidence instead of "hopes", since the community's thought
>> process is at stake here.
>> 
>> From the preliminary results we've seen, PIE has been able to address data
>> center issues quite well. There is no evidence thus far why PIE's control law
>> with auto-tuned parameters cannot adapt to data center or other network
>> environments. 
> 
> Taking what is mentioned in above sentence granted, there is no evidence if it
> cannot or if it can and this point is also orthogonal to my response to Anoop
> which was about his question at IETF about AQM deployment in general.


Just pointing out the evidence here --
http://www.ietf.org/proceedings/86/slides/slides-86-iccrg-5.pdf

The slides contain preliminary results from DC scenarios, showing PIE's
parameters and adaptability to DC scenarios. Thank god for the IETF archives
:)

Thanks,
Preethi