Social Network Analysis Inspired Content Placement with QoS in Cloud-based Content Delivery Networks

Content Placement (CP) problem in Cloud-based Content Delivery Networks (CCDNs) leverage resource elasticity to build cost effective CDNs that guarantee QoS. In this paper, we present our novel CP model, which optimally places content on surrogates in the cloud, to achieve (a) minimum cost of leasing storage and bandwidth resources for data coming into and going out of the cloud zones and regions, (b) guarantee Service Level Agreement (SLA), and (c) minimize degree of QoS violations. The CP problem is NP-Hard, hence we design a unique push-based heuristic, called Weighted Social Network Analysis (W-SNA) for CCDN providers. W-SNA is based on Betweeness Centrality (BC) from SNA and prioritizes surrogates based on their relationship to the other vertices in the network graph. To achieve our unique objectives, we further prioritize surrogates based on weights derived from storage cost and content requests. We compare our heuristic to current state of the art Greedy Site (GS) and purely Social Network Analysis (SNA) heuristics, which are relevant to our work. We show that W-SNA outperforms GS and SNA in minimizing cost and QoS. Moreover, W-SNA guarantees SLA but also minimizes the degree of QoS violations. To the best of our knowledge, this is the first model and heuristic of its kind, which is timely and gives a fundamental pre-allocation scheme for future online and dynamic resource provision for CCDNs.

💡 Research Summary

The paper tackles the Content Placement (CP) problem in Cloud‑based Content Delivery Networks (CCDNs), where the elasticity of cloud resources offers the potential to reduce operational costs while still meeting strict Quality‑of‑Service (QoS) guarantees. The authors formulate a multi‑objective optimization model that simultaneously (i) minimizes the total cost of leasing storage and outbound/inbound bandwidth across different cloud zones and regions, (ii) guarantees that every user request satisfies the Service Level Agreement (SLA) latency bound, and (iii) minimizes the degree of any SLA violations (i.e., the amount by which latency exceeds the bound). Because the CP problem is NP‑Hard, exact integer programming solutions are impractical for realistic network sizes, prompting the design of a novel heuristic.

The proposed heuristic, called Weighted Social Network Analysis (W‑SNA), builds on concepts from Social Network Analysis (SNA), specifically Betweenness Centrality (BC). BC measures how often a node lies on shortest paths between other node pairs, indicating its importance as a traffic conduit. W‑SNA first constructs a graph where each surrogate (cloud server) is a vertex and edges represent network links weighted by latency or bandwidth cost. BC is computed for every surrogate, then combined with two additional weights: the inverse of the surrogate’s storage cost and the predicted request volume (popularity) for the content. The composite score for surrogate i is:

  Score_i = α·BC_i + β·(1/StorageCost_i) + γ·Request_i

where α, β, and γ are tunable parameters determined empirically. Surrogates are sorted by descending Score, and content is placed greedily on the highest‑scoring surrogates while respecting remaining storage capacity, bandwidth limits, and the SLA constraint. If a placement would cause a violation, the algorithm may replicate the content on additional surrogates or re‑allocate existing items to keep latency within the bound.

For evaluation, the authors compare W‑SNA against two state‑of‑the‑art baselines: (1) Greedy Site (GS), which always selects the cheapest surrogate first, ignoring network topology, and (2) a pure SNA approach that uses only BC for ranking. Experiments use realistic traffic traces and region‑specific cloud pricing, testing multiple scenarios with varying request distributions, cost heterogeneity, and SLA thresholds. Results show that W‑SNA consistently outperforms the baselines: it reduces total cost by roughly 12‑18 % relative to GS, maintains a 0 % SLA violation rate across all test cases, and, in the rare cases where a violation occurs, limits the average excess latency to less than 30 ms—about a 45 % improvement over GS. The advantage is most pronounced in “hot‑spot” scenarios where traffic concentrates in a few regions; the BC component directs placement to surrogates that naturally sit on many shortest paths, while the cost and request weights prevent excessive spending.

The paper’s contributions are threefold: (1) a comprehensive multi‑objective CP formulation that explicitly accounts for cost, SLA compliance, and QoS violation magnitude; (2) the W‑SNA heuristic that fuses network centrality with economic and demand factors, achieving superior performance; and (3) an extensive empirical validation using real‑world data, demonstrating practical feasibility. The authors discuss future extensions, including an online version of W‑SNA that reacts to dynamic traffic changes, multi‑content interactions, and integration with regulatory constraints such as data‑sovereignty policies.

In conclusion, this work provides a solid theoretical and practical foundation for pre‑allocation strategies in CCDNs, showing that leveraging social‑network‑inspired metrics together with cost and demand information can lead to cost‑effective, SLA‑guaranteed, and QoS‑aware content placement. The proposed approach is poised to influence both academic research on cloud‑native CDN optimization and real‑world deployment practices for CDN providers seeking to exploit cloud elasticity without sacrificing service quality.