Shorter waiting times with new mathematics

Publication date
27 Apr 2009

Waiting in traffic, waiting for the ATM-machines, "There are more than ten people waiting before you": shorter waiting times in a queue, which person doesn’t want that? PhD Wemke van der Weij (CWI) developed new, highly efficient mathematical methods to shorten queues. With these methods the operating capacity will be divided between more customers. Van der Weij received her PhD-degree on April 23 at the Vrije Universiteit van Amsterdam on her thesis "Queuing Networks with Shared Resources”. She experimentally proved that her results lead to major improvements in the performance of queuing systems such as web servers.

Queues are everywhere. In recent decades these systems have been successfully studied assuming that the service capacity - how many customers are being served in a certain time – from each row is independent of the number of waiting customers in other rows. In many modern applications such as web servers, file servers, database servers, mobile networks and distributed software systems, this isn't the case. In these systems the total operating capacity is being divided dynamically among the different queues. This is called shared capacity. "A supermarket is a good example in order to clarify this, where a greengrocer or butcher can help the baker when it is very busy," says Van der Weij.

Mathematician Van der Weij studied different characteristics of queues with shared capacity. She developed new insights and strategies for optimal strategies for allocation of capacity to customers in queues where customers share control capacity. The results are examined in a laboratory and validated. These strategies led to significant improvements in the performance of the systems. Tests with web servers showed that the performance of heavily loaded servers could be greatly improved by the proper allocation of capacity to the different web users.

More information ~ Weij /
supervisor: Prof. Dr. R.D. van der Mei

The thesis is published by Ipskamp printers, ISBN: 978 90 8659 298 2.
The study is co-financed by the Bsik BRICKS program.



Meer informatie
promotor: prof. dr. R.D. van der Mei

Het proefschrift is uitgegeven bij Ipskamp drukkers, ISBN: 978 90 8659 298 2‏.
Het onderzoek is mede-gefinancierd door het Bsik-BRICKS programma.