PROJECT TITLE :

Predicting Transient Downtime in Virtual Server Systems: An Efficient Sample Path Randomization Approach

ABSTRACT:

A central challenge in developing cloud datacenters Service Level Agreements is the estimation of downtime distribution of a group of provisioned servers over a service window, which is compounded by three facts. Initial, while steady-state chances are derived for birth-death processes involving server failures and repairs, they could be highly inaccurate below transience. Furthermore, steady-state cannot be assured underneath typical service windows. Thus, estimation of transient distributions is crucial. Second, the processes of failures and repairs could follow any distribution and hence would like to be extracted using system log information and modeled using acceptable general distributions. Third, downtime distributions over service windows rely on the number of servers and their deployment structure for a contract. We have a tendency to develop an efficient and generalized sample path randomization approach to exactly estimate transient chances under three totally different checkpointing ways and 3 flexible failure distribution models. The estimators are unbiased, consistent, efficient and sufficient. Their asymptotic convergence is established. The estimation algorithms are computationally efficient in solving practical problems and yield wealthy information on transient system behaviors. The methodology is general and extensible to numerous server failure and repair processes characterised using birth-death modeling.