دانلود رایگان مقاله طراحی و ارزیابی مدل ذخیره سازی هماهنگ در شبکه برای شبکه محتوا محور

Abstract

In-network caching has been widely adopted in Content Centric Networking (CCN) to accelerate data delivery, mitigate server load and reduce network traffic. However, the line-speed requirement makes the in-network caching space very limited. With the rapid growth of network traffic, it is significant challenging to decide content placement in such limited cache space. To conquer this conflict, coordinated in-network caching schemes are needed so as to maximize the profit of ubiquitous caching capacities. In particular, in-network caching in CCN is deployed as an arbitrary network topology and naturally supports dynamic request routing. Therefore, content placement scheme and dynamic request routing are tightly coupled and should be addressed together. In this paper, we propose a coordinated in-network caching model to decide the optimal content placement and the shortest request routing path under constraints of cache space and link bandwidth in a systematic fashion. Via extensive simulations, the effectiveness and efficiency of our proposed model has been validated.

7. Conclusion

In this paper, we investigate the characteristics of arbitrary network topology and dynamic request routing requirement of CCN innetwork caching. In response to these characteristics, we propose a complete model to formulate the coupled content placement and request routing problem under link bandwidth and cache space constraints, aiming at maximizing the bandwidth cost savings. Experiments validate the effectiveness of the proposed model and show that content popularity, cache size allocation among routers, total cache size capacity and number of direct links are important factors and significantly affects the network performance. There are several directions for future works. First, as we mentioned earlier, the in-network caching system is highly dynamic. It is thus wise to investigate how the temporal dynamics impact the coordination of content placement and dynamic request routing. Second, our work replies on the length of each time slot and the setting of Lagrangian multiplier. Therefore, it is important yet challenging to analyze the impact of time slot division on the convergence of Lagrangian multiplier.