An Overview of Cloud Based Content Delivery Networks: Research Dimensions and State-of-the-Art - Transactions on Large-Scale-Data-and Knowledge-Centered Systems XX - page 134

Fig. 1. The architecture of a Content Delivery network

Table 1. CDN Request Redirecting mechanisms
Global Server Load Balancing Global awareness

Smart authoritative DNS

DNS-based request routing
HTTP redirection
URL rewriting URL modication

Automation through scripts

Anycasting IP anycast

Application level anycast

CDN Peering Centralized directory model

Distributed Hash Table
Flooded request model
Document routing model

for the next probable requirement; as it might be possible that the same/other user(s)
in the region will require the content. Prefetching [ 2 ] is another important func-
tionality provided by the POP server where it fetches the content that clients may be
interested in from the origin server thereby reducing the chance of traf c congestion

especially during the high demand. Needless to say, prefetching needs to predict the
users

preferential contents by synthesizing and analysing the historical information
such as access logs. It is evident that this kind of prefetching techniques may require
statistical data mining algorithms to determine what content to prefetch.
3. Request Redirecting mechanism: One of the functions of a CDN is to dynamically
redirect clients to the most optimal servers based on several QoS parameters such as
server load, latency, network congestion, client access networks, and proximity etc.
There are a variety of methods that can be used to implement this mechanism as
presented in Table 1 [ 3 ].

'