Content delivery in the internet is one of the most challenging technology issues partly because of the increasing numbers of end-users and partly because the available designs do not address the issues of locality of the end user. They also fail to address the cost associated with delivery of the content. This has seen quite a number of Internet Service Providers ISP operate hundreds or thousands of data servers in their center just to enable them to handle the fluctuating and sometimes torrential information traffic. The two articles address two distinct but related novel ways of handling and delivering traffic over the internet without the usual challenges of numerous servers or higher costs. This paper summarizes the two articles by evaluating the objective of each article, methods used in the analysis, assumptions, and the results revealed.
The first article introduces a new way of controlling and managing internet traffic called Content-aware Traffic Engineering (CaTE). Its objective is to propose a control concept that allows significant amount of content requested by different end users to effectively be served from different locations without interfering with the security or delivery protocols. In its results, the authors demonstrate that using the CaTE technology, an Internet Service Provider (ISP), can essentially influence the path of traffic going through their network. Another underlying fundamental of CaTE concept is the collaboration between Content Delivery Networks (CDN) and ISP. The authors assume that it can provide creative traffic engineering to give an end-user a new experience without necessarily going into the specifics of routing. This therefore marks a major difference with the common CDN-ISP collaboration that relies so much on routing to deliver content to the end-user (Poese et al, 22).
The authors of the article extensively use prototype CaTE systems to illustrate how the concept can scale up the many requests (probably, thousands per second) from end-users and how CaTE concept helps in handling such massive requests. Using larger tier-1 systems, the article demonstrates how improved performance on the side of the CDN and ISP can be achieved through CaTE concept. The main concern in the CaTE concept is to provide a topology that will allow deployment of CDN with ISP using the new design without affecting the delivery of data. CaTE concept is also advantageous because it is compatible with popular applications, including Netflix, which have started offering their CDN services with popular ISP. Results showed that CaTE would allow reduction in utilized links and also in Hypertext Transfer Protocol HTTP traffic, as they will shift towards the available shorter paths provided by CaTE technology. Inherent challenges such as a reduction in traffic, having a correspondent increase in utilization of higher load link, can be addressed through careful deployment of CaTE inside a network (Poese et al. 25). A number of benefits are lined up when the new concept is implemented including reduction of costs of acquiring and maintaining thousands of servers by major ISP and CDN, and also the end-user will have the benefit of accessing content at a faster and more efficient manner.
The second article also addresses the same topic of content delivery but proposes a new approach thereto; the fundamental principle in the new traffic design is called ISP-centric content delivery (iCODE). Through iCODE, it is possible for routers or network entities to exploit the in-network storage modules for content caching and content delivery. Some of the assumptions in this system design are that the physical network exhibits an inherent autonomy. This is based on the fact that it is operated by only one ISP that must have many routers, referred to as content routers, together with storage modules, mainly for caching of the content. The design is based on the principle that iCODE knowledge of popular cached content. That content will be redirected to the C-routers, thus allowing similar content to be serviced within other routers within the same ISP network. Results show that a major advantage of this approach in content delivery is that the moment a popularity of some given content is established within the system, the same content is cached and stored in multiple C-routers so that subsequent requests can be received from whichever router receives the request from the next end-user. This will eliminate the operation of the current ISP, where requests for content can be passed through several C-routers before the correct router is found, which in itself has proved to be costly and time-consuming (Cho et al, 157).
The two articles propose the advantages of CDN and ISP by adapting the proposed new ways—the CaTE and the iCODE, which includes achieving stable and reduced latency during data transfer and thus giving the user a unique experience in the data transfer. The two articles also provide unprecedented novelty in traffic engineering and topology in content delivery using cached content and spatiality of tempo ability of the request content from the end-user. Furthermore, iCODE provides the ISP with several options to allow incremental deployment to allow backward compatibility with the current CDN. It also allows for the provision of independent services from a particular ISP different from other ISP. This gives the end-user a unique experience whenever he switches from one ISP to the other. A common feature between the two proposed content delivery systems is the enhanced privacy and performance mechanisms that do not allow the revelation of sensitive information being communicated between the end-user and the servers. In retrospect, CDNs in both designs only allow the identity of the server to the end-user without allowing for additional information such as server load or the costs associated with the delivery of that information. The two thus eliminate the present ISP delivery models that allow public information on ISP network data, routing weights, and ranking scores.