[Bloat] ipspace.net: "QUEUING MECHANISMS IN MODERN SWITCHES", > (Jonathan Morton)

[Neil Davies]

On 28 May 2014, at 18:29, David Collier-Brown <davec-b at rogers.com> wrote:

> On 05/28/2014 11:33 AM, Jonathan Morton <chromatix99 at gmail.com> wrote
>> It's a mathematical truth for any topology that you can reduce to a black box with one or more inputs and one output, which you call a "queue" and which *does
> not discard* packets.  Non-discarding queues don't exist in the real
> world, of course.
>> 
>> The intuitive proof is that every time you promote a packet to be transmitted earlier, you must demote one to be transmitted later.  A non-FIFO queue tends to increase the maximum delay and decrease the minimum delay, but the average delay will remain constant.
> 
> A niggle: people working in queuing theory* make the simplifying
> assumption that queues don't drop. When describing the real world, they
> talk of "defections", the scenario where a human arrives at the tail of
> the queue and "defects", either to another queue or to the exit door of
> the store!

There is another mathematical approach that we've found very useful, actually the original work goes back to the 1950's (M/M/1/K/K). 

As mentioned in a reply just now in a different thread, it does give some interesting insights into the underlying two-degrees of freedom that are present in every finite queue. 


> As you might guess, what I find intuitive the IP world finds wrong, and
> vice versa.
> 
> --dave
> [* as opposed, perhaps, to queuing networks (:-)]
> -- 
> David Collier-Brown,         | Always do right. This will gratify
> System Programmer and Author | some people and astonish the rest
> davecb at spamcop.net           |                      -- Mark Twain
> 
> 
> 
> -- 
> David Collier-Brown,         | Always do right. This will gratify
> System Programmer and Author | some people and astonish the rest
> davecb at spamcop.net           |                      -- Mark Twain
> _______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat