Murray S. Kucherawy wrote:
I don't think the practice of connection caching is particularly
selfish when compared to the cost of having the connection torn down and
then re-established with some frequency, when it's generally much cheaper
for both the sender and the receiver to just leave it open.
Only if there is efficient work being done. Look, maybe it was my
fault to assume the obviousness of this classic Operational Research
and Querying Design problem is commonly understood.
One just need to use Little's Law [1] to see how this is a very
selfish queuing concept that benefits only one part of the system and
increases demands and cost on other parts. This is particular true
you reach a capacity and can no longer remain in steady state. To
return it back to steady state, adjustments are needed. The flaw in
CS is the erroneous presumption that there is unlimited resources on
the receiver side (service workers) so that the waste in idle service
time is factored out as arrival rates increased. As in a bank
scenario, efficiency is lost if after a teller station finishes
servicing a customer, he decides to hangs around the station doing
nothing. The teller's time is wasted. Less workers available, Queued
Customers wait longer, Lines get longer. Little's Law will show you this.
[1] http://en.wikipedia.org/wiki/Little's_law
--
Sincerely
Hector Santos
http://www.santronics.com