Investigation and Analysis of Google Cluster Usage Traces: Facts and Real-Time Issues

Abstract

The term "Cloud" has become ubiquitous in our life since it deals with most of our activities. In fact, we are practising cloud services in most of our social, academic, and business tasks. The technological era we are currently living in, is almost based on Internet services. Nowadays, cloud computing and cloud load balancing are considered as the core for cloud services. Thus, the fields of cloud computing and cloud load balancing have attracted various researchers to develop innovative methods aiming at enhancing cloud services and providing cloud end users with adequate services. This paper discusses the field of cloud load balancing in two important aspects. The first aspect includes taking a deep look on the tasks and servers on Google cluster usage traces, while the second aspect includes the discussion about how load balancing algorithms deal with tasks that have time constraints (e.g., deadline). The findings of this paper exhibited that when using real time concepts with the selected scheduling algorithms, such algorithms require further enhancements to satisfy the need of real time tasks. Netlogo simulator was used as a tool to execute the practical side of the work.