Elucidation of Upcoming Traffic Problems in Cloud Computing

Abstract

Cloud computing is generally believed to the most gifted technological revolution in computing and it will soon become an industry standard. It is believed that cloud will replace the traditional office setup. However a big question mark exists over the network performance when the cloud traffic explodes. We call it “explosion” as in future we know that various cloud services replacing desktop computing will be accessed via cloud and the traffic increases exponentially. This paper aims at addressing some of these doubts better called “dangers” about the network performance, when cloud becomes a standard globally. Our study concentrates on, that despite of offering better round-trip times and throughputs, cloud appears to consistently lose large amounts of the data that it is required to send to the clients. In this paper, we first give a concise survey on the research efforts in this area. Our survey findings show that the networking research community has converged to the common understanding that a measurement infrastructure is insufficient for the optimal operation and future growth of the cloud. Despite many proposals on building an network measurement infrastructure from the research community, we believe that it will not be in the near future for such an infrastructure to be fully deployed and operational, due to both the scale and the complexity of the network. We explore this problem, and offer deduction that might explain this erratic behavior. We also suggest a set of technologies to identify and manage cloud traffic using IP header DS field, QoS protocols and some high speed edge routers. Our solutions assume that cloud is being assessed via basic public network.