Cost-Aware Cloud Bursting for Enterprise Applications

摘要: The high cost of provisioning resources to meet peak application demands has led the widespread adoption pay-as-you-go cloud computing services handle workload fluctuations. Some enterprises with existing IT infrastructure employ a hybrid model where enterprise uses its own private for majority computing, but then “bursts” into when local are insufficient. However, current commercial tools rely heavily on system administrator’s knowledge answer key questions such as burst is needed and which applications must be moved cloud. In this article, we describe Seagull, designed facilitate bursting by determining should transitioned automating movement process at proper time. Seagull optimizes using an optimization algorithm well more efficient approximate greedy heuristic. also overhead deploying intelligent precopying mechanism that proactively replicates virtualized applications, lowering time from hours minutes. Our evaluation shows over 100p improvement compared naive solutions produces expensive ILP. scalability our dramatically better number VMs increase. illustrates scenarios prototype can reduce costs than 45p cloud, incremental added offset reduction nearly 95p.