Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!decvax!harpo!eagle!mhuxl!ihnp4!inuxc!pur-ee!uiucdcs!uicsl!mather
From: mather@uicsl.UUCP
Newsgroups: net.flame
Subject: Re: Supermarket Flames! - (nf)
Message-ID: <5543@uiucdcs.UUCP>
Date: Sat, 11-Feb-84 22:44:34 EST
Article-I.D.: uiucdcs.5543
Posted: Sat Feb 11 22:44:34 1984
Date-Received: Wed, 15-Feb-84 03:53:58 EST
Lines: 28

#R:ihu1g:-17000:uicsl:4300147:000:1142
uicsl!mather    Feb 10 11:34:00 1984

What would be wrong with a queue structure similar to that of the post
office ? Single stream, multi-processor system. That is, have one long
line with an n-way branch to the cashiers. The next available cashier
grabs the next item off the queue (me!)

It would avoid the problem of a cashier getting off break, walking up to a
closed register and saying, "I can take people over here" and having a mad
dash (from the tails of the other queues) to that register.

I realize that this ignores the problem of those that only want to buy a
bag of potato chips and must have to endure a long line, but isn't
first-come-first-serve (FIFO) THE AMERICAN WAY?

When I'm at the post office, I really like that system. It seems fairer.

What is the probability distribution for 'response time' for this model,
given that the input to the queue is Poisson and the cashiers handle each
queue item (the people) with different Poission rates lambda-sub-i ?

				B.C.Mather (I always get the slow line!)
				Le Maitre
				...!uiucdcs!uicsl!mather

I wonder if this 'long queue' would be shaped like the cattle stalls that
the adventure parks (Disney) have?