Computer Science Assistant Professor publishes as first author in IEEE Transactions in Cloud Computing, a 7.9 impact factor journal

Computer science assistant professor Chenxi Qiu, Ph.D. had his paper entitled “Dynamic Demand Prediction and Allocation in Cloud Service Brokerage” accepted by, IEEE Transactions in Cloud Computing, a premier journal with an impact factor of 7.928.

A cloud service brokerage (CSB) is a third-party individual or business that acts as an intermediary between cloud providers and tenants. A CSB reserves the cloud resources (e.g., servers) from the cloud providers and sells services (e.g. virtual machine (VM)) along with administration and security to the tenants with higher prices. Usually, CSBs can make cloud services more valuable for cloud tenants, because CSBs work closely with cloud providers to get price breaks or access to more information about how much resources are required for a service. In addition, CSBs can enhance the security of cloud services for tenants because they can monitor, track, protect and enforce company policies across all demands from different tenants. Thus, CSBs can make it easier, less expensive, safer and more productive for tenants to use cloud resources, particularly when a tenant’s demands span multiple cloud service providers.

To maximize its own profit, CSBs usually try to distribute tenant demands to reserved servers such that the total reservation cost is minimized with the tenants’ service level agreement (SLA) being satisfied. The demand allocation problem for CSB is non-trivial to solve due to uncertainty of tenants’ behavior. To avoid possible violations among demands, existing schemes allocate additional padding resources on the predicted demands, which leads to under-utilization of reserved resources.

"Accordingly, we proposed a Probabilistic Demand Allocation (PDA) system to address the demand allocation problem for CSB. In PDA, we not only predict tenants’ demands based on their historical records, but also estimate the probability distribution of prediction errors. As over- and under-estimation are equally likely to happen with our prediction method, when allocating demands to a single server, their errors are possibly offset. Hence, it is unnecessary to allocate additional resource to each demand for violation prevention. Given the predication results, we applied math optimization techniques (e.g., stochastic optimization) for demand allocation, of which the objective is to minimize the overall cost of reserved servers while satisfying tenants’ SLA with high probability. Finally, we demonstrated the superiority of PDA compared with state of the arts with both trace-driven simulation and real-world experiment," explained Qiu.