2016/12/06 by Brendan Patch, Patch, Brendan, Thomas Taimre +1
Business, Management and Accounting · Computer Science · Social Sciences · #Advanced Queuing Theory Analysis #Cloud Computing and Resource Management #Distributed #FOS: Computer and information sciences #Parallel #Transportation Planning and Optimization #and Cluster Computing (cs.DC)
paper · pdf · doi:10.48550/arxiv.1612.01845
openalex publication_date 2016/12/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
User demand on the computational resources of cloud computing platforms\nvaries over time. These variations in demand can be predictable or\nunpredictable, resulting in `bursty' fluctuations in demand. Furthermore,\ndemand can arrive in batches, and users whose demands are not met can be\nimpatient. We demonstrate how to compute the expected revenue loss over a\nfinite time horizon in the presence of all these model characteristics through\nthe use of matrix analytic methods. We then illustrate how to use this\nknowledge to make frequent short term provisioning decisions --- transient\nprovisioning. It is seen that taking each of the characteristics of fluctuating\nuser demand (predictable, unpredictable, batchy) into account can result in a\nsubstantial reduction of losses. Moreover, our transient provisioning framework\nallows for a wide variety of system behaviors to be modeled and gives simple\nexpressions for expected revenue loss which are straightforward to evaluate\nnumerically.\n