Abstract
We consider a website host server with web quality of service (QoS) capabilities to offer differentiated services. A quantitative modelling framework is set up to analyse the economic benefits of differentiated services and to build optimization models for managing the website host's connection bandwidth to the Internet (which is assumed to be the bottleneck factor determining the QoS). Three models are formulated corresponding to three operational scenarios to provide differentiated services. The first is for the marketing manager to classify visit requests as premium or basic when the information technology (IT) manager has already reserved bandwidths for the two classes, and the second is for the IT manager to allocate the total available bandwidth to each class when the marketing manager has already designated which visit requests are premium and which are basic. The third is for the joint optimization of request classification and bandwidth allocation when centralized coordination is possible. Analytic results are obtained for a special case that corresponds to very impatient customers requesting large amounts of data. Qualitative insights gained and numerical results obtained strongly support the implementation of differentiated services. More interestingly, the decentralized models that use simple and rough-cut rules yield solutions almost as good as the joint optimization model.
Similar content being viewed by others
References
Abramowitz M and Stegun IA (1965). Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Dover: New York.
Bayles DL (2001). E-Commerce Logistics and Fulfillment: Delivering the Goods. Prentice-Hall: Upper Saddle River, NJ.
Bernet Y et al (2005). A Framework for End-to-End QoS Combining RSVP/Intserv and Differentiated Services, www.ietf.org/proceedings/98dec/I-D/draft-bernet-intdiff-00.txt, accessed 22 April 2005.
Bhatti N and Friedrich R (1999). Web server support for tiered services. IEEE Network 13: 64–71.
Chassioti E and Worthington DJ (2004). A new model for call centre queue management. J Opl Res Soc 55: 1352–1357.
Gibbens RJ and Kelly FP (1999). Resource pricing and the evolution of congestion control. Automatica 35: 1969–1985.
Haskose A, Kingsman BG and Worthington D (2002). Modelling flow and jobbing shops as a queueing network for workload control. Internat J Prod Econom 78: 271–285.
Heyman DP and Sobel MJ (1982). Stochastic Models in Operations Research, Volume I: Stochastic Processes and Operating Characteristics. McGraw-Hill: New York.
Kehoe CM et al (2005). Results of GVU's Tenth World Wide Web User Survey, www.gvu.gatech.edu/user_surveys/survey-1998-10/tenthreport.html, accessed 22 April 2005.
Kelly FP, Mauloo AK and Tan DKH (1998). Rate control for communication networks: Shadow prices, proportional fairness and stability. J Opl Res Soc 49: 237–252.
Key P and McAuley D (1999). Differentiated QoS and pricing in networks: where flow control meets game theory. IEE Proc Software 146: 30–43.
Nielsen J (2005). The Need for Speed, www.useit.com/alertbox/9703a.html, accessed 22 April 2005.
Ramsay J, Barbesi A and Preece J (1998). A psychological investigation of long retrieval times on the World Wide Web. Interacting Comput 10: 77–86.
Shanthikumar JG and Xu SH (1997). Asymptotically optimal routing and service rate allocation in a multiserver queueing system. Opns Res 45: 464–469.
Tang C and Yoo SC (1991). System planning and configuration problems for optimal system design. Eur J Opl Res 54: 163–175.
Van Laarhoven PJM and Aarts EHL (1987). Simulated Annealing: Theory and Applications. D Reidel: Dordrecht.
Whitt W (1999). Partitioning customers into service groups. Mngt Sci 45: 1579–1592.
Acknowledgements
We thank two anonymous referees for their suggestions that have improved the paper. The research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC).
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A. Proof that expected speed G(λ) is decreasing in λ
Since z=λ/D0, it is sufficient to prove that G(z) is decreasing in z. We can write the expected speed in a slightly different form as
Differentiating we find
where
Since the denominator of (A1) is always positive, we attempt to show that the numerator ν(z) is always negative in order to prove that G(z) is decreasing in z. We write
Taking m+n−1=k, this can be re-written as
Now, if
then . However,
From this, it now follows that if k is even, then grouping terms as above, we obtain and hence v(z)<0 which implies G′(z)<0.
If k is odd, let k=2s+1. Then, except the middle term in all other terms sum to a negative number. Since the middle term is
the lemma is proven. □
Appendix B. Proof that expected speed G(b) is increasing in b
First, we note that
Since
differentiating G(b) with respect to b, we find
where the numerator ν(b) is
with
It is easy to re-write the numerator as, . It is now possible to simplify the coefficient of zu by grouping terms as follows:
Without loss of generality, we can assume m⩽u/2. Note that m=u/2 is possible only when u is even. Now, when m<u/2 we note that
So, Cm,u−m+Cu−m,m>0
Now, when u is even, we also need to consider the middle term in its coefficient, which is,
Thus, all the terms in the coefficient of zu are positive, which proves the lemma. □
Rights and permissions
About this article
Cite this article
Ou, J., Parlar, M. & Sharafali, M. A differentiated service scheme to optimize website revenues. J Oper Res Soc 57, 1323–1340 (2006). https://doi.org/10.1057/palgrave.jors.2602100
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1057/palgrave.jors.2602100