Cloud computing is changing our lives in many ways. Any user adopting cloud computing services certainly expect the kind of improved performance that an cloud computing environment should provide. Workload is one of the major factor to accomplish high performance on Clouds. Workload classification would be a good solution to improve the performance. Analysis and Classification of the workload in cloud computing is challenging due to the virtualization layer overhead, complexity in workloads and insufficient availability of datacenter tracelogs. These elements adds lack of methodologies to characterize the applications hosted in the cloud. The workload can be either the synthetic or genuine workload. Google clusters traces, Yahoo M45 Hadoop cluster, PlanetLab cloud traces are some of the genuine cloud tracelogs available. Understanding characteristics of workload in cloud will help both cloud suppliers and researchers. Cloud suppliers can enhance their system Quality of Service (QOS) and researchers can assess new approaches using cloud simulators like CloudSim, CloudAnalyst, GreenCloud etc. For better resources management Workload must be classified as simple as possible. From workload classification, performance models can be constructed such as energy efficiency and resource management. The main objectives of this paper is to present an idea about why workload is important, why workload classification is important, complexities found in classifying the workloads in heterogeneous cloud environment, review of some methods for analysis and classification of workload, and a proposed method for classification.
No. of Downloads :