INDUSTRIAL ENGINEERING AND OPERATIONS RESEARCH

PRESENTS

IEOR MONDAY SEMINAR

 

FEBRUARY 6, 2006

 

 

State Space Collapse in Many-Server Diffusion Limits of Parallel Server Systems

 

Tolga Tezcan

 

School of Industrial and Systems Engineering

Georgia Institute of Technology

www.isye.gatech.edu/_ttezcan

 

 

Abstract

 

We consider a class of queueing systems that consist of server pools in parallel and multiple customer classes. Customer service times are assumed to be exponentially distributed. We study the asymptotic behavior of these queueing systems in a heavy traffic regime that is known as the

Halfin and Whitt many-server asymptotic regime where the number of servers and arrival rates grow to infinity in a certain manner. Specifically, we extend the state space collapse framework of Bramson [1] to show that a class of state space collapse results in the Halfin and Whitt many-server diffusion limits of a parallel server system hold if the hydrodynamic limits of the same system satisfy certain conditions. We also show that the hydrodynamic limits satisfy a set of deterministic equations known as the hydrodynamic model equations that are related to but different from the fluid model equations.

 

We illustrate the application of our results in an inverted-V parallel server system that consists of multiple server pools and a single customer class. We propose the load-balancing (LB) routing policy to balance the utilizations of all the servers in the system with no unnecessary idling. We show that this policy balances both the long-run and finite-time average utilizations over all the server pools in the Halfin and Whitt regime.

 

 

 

TIME AND LOCATION: 3:30 - 4:30 P.M. - 3108 ETCHEVERRY HALL

 REFRESHMENTS WILL BE SERVED @ 3:00 P.M.