Cost Optimization for Dynamic Replication and Migration of Data in Cloud Data Centers

ABSTRACT:

Cloud storage Providers (CSPs) offer geographically information stores giving a few storage classes various costs. An essential issue looking by cloud clients is the way to misuse these capacity classes to serve an application with a period fluctuating workload on its articles at minimum cost. This cost comprises of private cost (i.e., capacity, Put and Get expenses) and potential relocation cost (i.e., arrange cost). To address this issue, we initially propose the ideal disconnected calculation that use dynamic and direct programming methods with the suspicion of accessible correct learning of workload on objects.

Because of the high time intricacy of this calculation and its prerequisite for from the earlier information, we propose two online calculations that make an exchange off amongst private and movement costs and powerfully select capacity classes crosswise over CSPs. The main online calculation is deterministic with no need of any information of workload and causes close to 2 1 times of the base cost acquired by the ideal disconnected calculation, where is the proportion of the private cost in the most costly information store to the least expensive one in either system or capacity cost. The second online calculation is randomized that influences “Residing Horizon Control” (RHC) procedure with the abuse of avalaible future workload data for w availabilities. This calculation causes at most 1 + w times the ideal cost. The viability of the proposed calculations is shown through recreations utilizing a workload blended in view of qualities of the Facebook workload.

LEAVE A REPLY

Please enter your comment!
Please enter your name here