You can add items from the left menu by dragging them here.

The ever-evolving nature of the Internet as a ubiquitous media for sharing, managing and delivering content and services gave birth
to Content Delivery Networks (CDNs) [1]. A CDN (Figure 1) is a collaborative collection of network elements spanning the Internet,
where content is replicated over several mirrored Web servers (i. e. surrogate) located at the edge of the network to
which end-users are connected. firewall software Content is replicated by the CDN's content distributor, either on-demand or beforehand, by
pushing the content from the origin server to the surrogates. As an end-user is served with the desired content from
the nearby server, the user ends up unknowingly communicating with a replicated surrogate close to him/her and retrieves content from
that server.

Over the last decade, considerable research efforts and momentum have been directed into this sphere, both from
the academia and the commercial developers. It could undoubtedly be considered as one of the top emerging technologies that will
have a major impact on the quality of science and society over the next 20 years. Having said that a
glimpse of the technological trends and future directions in this domain would be helpful to position researchers and practitioners at
the forefront of the field. In this article, we lay out a comprehensive research roadmap to the CDN community by
identifying potential research directions and technologies that will drive innovations within this domain.

A walk through CDN's evolution,as shown in Figure 2, reveals a smooth transition from pre-evolution period to the current generation of CDNs. It is
evident from the figure that the pre-evolutionary period has seen amended server deployment with improved caching techniques. As CDNs were
introduced, there has been a major boost in the market for delivering broadband content, streaming rich media contentvideo, audio and
associated dataover the Internet. Hence, several initiatives kicked off for developing hi-tech infrastructure for CDNs.

First generation CDNsmainly focused on static and dynamic content delivery. The main technological feature for this generation is the development of techniques
for intelligent routing, replication, and edge computing (splitting application and data across the origin and the edge). For the second
generation CDNs, the focus has shifted to Video-on-Demand (VoD), news on-demand, audio and video streaming with high user interactivity. This
generation has also witnessed the growth of CDNs (e. g. Ortiva Wireless) that are dedicated for content delivery to mobile
devices firewall hardware. This generation can be technologically attributed for the use of Peer-to-Peer (P2P), Cloud Computing, and energy-aware
techniques for content delivery and management. Although a few commercial CDNs provide capabilities of the second generation CDNs, most of
the efforts on this type of CDNs are still in their infancy and have not yet fully reached the market
[2][3]. It is anticipated that the third generation CDNs will be community-based CDNs, i. e. they will be mainly driven
by the common -people- or the average end-users. The main technological features for this generation of CDNs will be autonomic
(self-configuring, self-organizing, self-managing, and self-adapting) composition and content delivery, with the main focus on meeting Quality of Experience (QoE) of
end-users.

Dealing with the ongoing trends, CDN community is anticipated to experience a trend shift through new innovations. In
this context, an indicative list could be populated, realizing the awaited technological evolutions in the coming years.

-Recent innovations: With the introduction of the notion of -cloud computing-, the CDN community has experienced the growth of commercial
computing and storage delivery systems such as Nirvanix, Amazon S3, and Amazon EC2. They aim at delivering services that are
not only limited to Web applications, but also include storage, raw computing, or access to any number of specialized services.
In addition, researchers have come up with recent innovations such as P4P [11], portal for P2P to communicate with network
providers, and new traffic engineering models for cooperative content distribution [12]. These recent innovations utter for the fundamental issue in
current network management and control, and assist toward a systematic understanding and practical realization of the interactions between providers those
who generate and distribute content and those who provide an operation network resources. In future, potential research could focus on
identifying necessary application requirements, enhancing scalability, system robustness, usability and access performance, low cost, data durability, and support for security
and privacy.

- Cloud-based content delivery: Extension of traditional CDNs model to Cloud-based CDN or Content Delivery Clouds enhance
capabilities to deliver services that are not only limited to Web applications, but also include storage, raw computing or access
to any number of specialized services. The rising of Content Delivery Clouds call for a systematic understanding and practical realization
of the fundamental issues in current network management and control. In addition, it initiates potential research that focuses on identifying
necessary application requirements, enhancing scalability, system robustness, usability and access performance, low cost, data durability, and support for security and
privacy. MetaCDN (www.metacdn.com) [14][15] is an example system in this category providing flexible and adaptable content delivery services by harnessing
the state-of-the-art availability and reliability features, and security measures of upstream storage cloud providers with little overhead due to load
redirection.

- Scalability of dynamic content delivery: In recent years, industry efforts such as EdgeSuite content distribution from Akamai
and IBM WebSphere edge services have emerged to provide usage-based application and (dynamic) content delivery. However, the effectiveness of the
proposed solutions is still highly dependent on the access patterns of applications network security. In particular, the risk of
creating a bottleneck in the back-end layer is still one of the main issues that hinder the scalability of dynamic
Web content delivery. Moreover, the shifting of Web 2. 0 towards personalized (dynamic and uncachable) content, the presence of sensitive
personal user profile information, and convergence of Web 2. 0 with user mobility would disrupt content delivery services due to
the migration of users among edge nodes. Therefore, initiatives coping with this evolution could be foreseen from the CDN providers
and researchers studying solutions for dynamic Web content delivery.

- Applicability of anycasting: Although examined early in the CDN
evolution process, IP anycasting was not considered as a viable approach for request-redirection in CDNs. This is due to the
lack of load awareness and unwanted side effects of Internet routing changes on the IP anycasting mechanism. Nevertheless, the emergence
of route control mechanisms coupled with external intelligence to allow dynamic route selections [8] and recent Anycast based measurement work
[9], shed light on the possibility to realize CDN redirection using IP Anycast. Lately, a load-aware IP Anycast CDN architecture
[7] has been devised to show the applicability and usage of IP Anycast to facilitate proximal routing. These findings could
arouse interest in the CDN community to explore this area and to conduct research to gain operational experience in real
CDN deployments.

- Active content: The rapid growth of Peer-to-Peer (P2P) techniques and the improvement of digital content production
and retrieval tools may lead to the proliferation of user-generated active content. This type of content will be generated with
a great deal of autonomy to give it the self-awareness feature. Subject to autonomic replication, it can be wrapped with
an incentive-based lightweight cooperative agent with forecasting logic to predict end-user demand. Active content assures to retain its face-value, however,
can have different context-dependent meaning and interpretations. It can be represented with latent meta-data where context-oriented properties can be added
or removed during replication. However, it is necessary to maintain the provenience information along with the production and transformation of
active content so that the master copy can always be retrieved from the origin. Research in this context may draw
inspiration from several technological paradigms such as Web crawling, agent systems, economic principals, and social science. For this purpose, research
related to autonomic content delivery, such as autonomic replication of Web applicationsand policy-based content delivery in an active network can
be useful.

- Integration of media streaming: Hosting of on-demand media streaming service is challenging because of the enormous
network bandwidth required to simultaneously deliver large amount of content to end-users. To avoid network congestions and to improve performance,
P2P techniques can be used to extend the CDN peering system as an adaptive CDN. In such a system, content
storage and workload from streaming server, network and storage resources are offloaded to the end-users workstations. The fundamental idea is
to allow multiple subscribing peers to serve streams of the same video content simultaneously to a consuming peer rather than
the traditional single server-to-client streaming model, while allowing each peer to store only a small portion of the content. In
this regard, prior solutions such as cost-effective media streaming and adaptive streaming CDN can be useful.

- Mobility for
content delivery: Mobile networks are increasingly becoming popular for distributing information to a large number of highly dynamic users. In
comparison to wired networks, mobile networks are distinguished by potentially much higher variability in demand due to the user mobility.
Content delivery in CDN peering can take into account this mobility notion, by potentially considering very high spatial and temporal
demand variations to dynamically reconfigure the system, and to minimize the total traffic over the network backbone. A mobility model
in this context can be designed to allow the access of accurate and up-to-date information and enterprise applications.

-
Peering of CDNs: Present trends in content networks and content networking capabilities give rise to the interest in interconnecting CDNs
[10]. Cooperation between CDNs could reduce costs with over-provisioning and provide users with high quality services. Although appealing, there are
a number of challenges, both technical and non-technical (i.e. commercial and legal), that could hinder its rapid growth. As for
instance, the necessity of a common interaction protocol, the proprietary nature of a CDN to gain competitive advantage in the
market, and complex legal issues involved (e.g. embargoed or copyrighted content) may block off the nascence of peering CDNs in
the commercial domain. These challenges must be overcome to promote CDN peering. Future research is expected to be heavily focused
on such collaborations, including short-term peering wherein CDNs operate to handle flash crowds, or long-term peering in which they explore
the delivery of specialized services. In particular, an ad-hoc or planned peering of CDNs would require fundamental research to be
undertaken to address the core problems of measuring and disseminating load information, performing request assignment and redirection [13], enabling content
replication and determining appropriate compensation among participant CDN providers on a geographically distributed -Internet- scale.

- Deployment ofmarket mechanisms: An economic model exploiting the dynamism of the CDN market could be envisioned to make the system more
manageable through emergent marketplace behavior. This market drift could be driven based on Service Oriented Architectures (SOA) in the business
domain. In addition, new robust and scalable replication, resource sharing and load balancing techniques could come into the picture, guided
by utility functions that satisfy Quality of Service (QoS) requirements of end-users. Recent research [4][5][6] justify this view by plying
models using mathematical principles, game theory and business practices to reflect the pricing dynamics in current CDN market.

In addition to the above listed future research directions, we can anticipate the integrated use of existing emerging as well
as stable technologies, e.g. agent, P2P, grid, data mining, to augment the effectiveness and boost the efficiency of future CDN
infrastructures. We also perceive a possible shift in the CDN industry as CDN internetworking, adaptive CDNs, mobile CDNs, and to
the full, community-based CDNs are evolving. Finally, the recent consolidations in the CDN industry due to acquisitions and/or mergers, may
lead to new issues in the design, architecture, and development of CDNs. We hope that readers would get a better
understanding and interpretation of essential concepts in this area by analyzing the present trends and future research directions in the
CDN domain, as stated in this article.

[2] Aioffi, W. M., Mateus, G. R., Almeida, J. M., andLoureiro, A. A. F. Dynamic content distribution for mobile enterprise networks. IEEE Journal on Selected Areas on Communication, 23(10), 2005.



[4] Christin, N. and Chuang, J. A cost-based analysis of overlay routing geometries. In Proc. of 24thAnnual Joint Conference of the IEEE Computer and Communication Societies (INFOCOM'05), 4, pp. 2566-2577, 2005.

[6] Kaya, C.,Dogan, K., and Mookerjee, V. An economic and operational analysis of the market of content distribution services. In Proc. of
the International Conference on Information Systems, 2003.


[8] Van der Merwe, J. E. et al. Dynamic connectivitymanagement with an intelligent route service control point. In Proc. of ACM SIGCOMM INM, 2006.

[13] Pathan andR. Buyya, "Resource discovery and request-redirection for dynamic load sharing in multi-provider peering content delivery networks," Journal of Network and
Computer Applications, vol. 32, no. 5, pp. 976-990, 2009.

[14] Pathan, M.,Broberg, J., and Buyya, R. "Maximizing utility for
content delivery clouds," Lecture Notes in Computer Science, Proc. 10th International Conference on Web Information Systems Engineering (WISE'09), vol. 5802,
pp. 13-28, 2009.

[15] Broberg, J., Buyya, R. and Tari, Z. "MetaCDN: Harnessing -Storage Clouds' for high performance content
delivery," Journal of Network and Computer Applications, vol. 32, no. 5, pp. 1012-1022, 2009.


In this context,
researchers in the CLOUDS Laboratory of the University of Melbourne, Australia are actively engaged in the development of next generation
user-level CDNs, under the flagship project CLOUDBUS. Please check: