Dimensioning the Cloud

In this thesis we use mathematical models to dimension clouds. We focus on loss models from the theory of queueing. We use the blocking probability in these models as performance measure of a cloud. We consider both infinite and finite source models as well as single-rate and multi-rate models for cloud dimensioning based on request blocking probability. These models enable providers to dimension both public and private clouds as well as clouds with one customer type and clouds with different customer types. In addition to surveying existing formulas, we also derive new formulas that can be used in practice. We also compare the different models and give recommendations about which model should be used in different cloud situations. Finally, we give a first quantification of the possible benefits of intercloud, the cooperation between cloud providers.

Leiden University Student Repository

Documents

In Collections

Dimensioning the Cloud