Content delivery networks: Insights and recent advancement

Network congestion, delay in response are the main hindrance for today's Internet based world. Content delivery network (CDN) was evolved to solve the issue by delivering content to the internet edge, closest to the user proximity. CDN was introduced to prevent flash crowd which is an unexpected upsurge in the network traffic [1]. CDN copies the content of the origin server to a number of cache servers over the network. The cache servers are named as surrogate servers or replica servers or edge servers. Theoretically these edge servers are deployed at the internet edges but practically they are placed at different geographically dispersed strategic locations of the network with the mission of reducing latency and bandwidth consumption. The performance of content delivery networks heavily depends on the decision of how many edge servers will be placed and where. This paper will discuss several replica server placement algorithms introduced by different authors. The placement algorithms include Tree-Based, Greedy, Max fan out, Flow count, HotZone, HotSpot, GeoIP and NetClust. An analysis among algorithms is done based on the optimization factors. This paper will indicate that Greedy algorithm is the best among all other heuristic approaches and it can be used in capacity constrained surrogate placement problem also. The CDN architecture is also shown along with the method through which the CDN delivers its content.