BASIC INFORMATION
Short name: UNIC
Long name: Uni kernel-based C DNs for 5G Networks
Company: University of Macedonia | www.uom.gr
Country: Greece
Call: F4Fp-04-M (see call details)
Proposal number: F4Fp-04-M27
SUMMARY REMARKS & TESTBEDS
The exponential growth of Internet content, in size, quantity and network traffic demands, enabled new network architectures realizing efficient hosting, discovery and dissemination of content, such as the Content Delivery Networks (CDNs). CDNs are usually based on large data centers, proprietary software and may not be responsive enough to dynamic changes in network conditions and user requirements. Such an approach is becoming inefficient for 5G networks targeting ultra-low latency services through lightweight edge clouds that host (or cache) the content near the end-users.
Along these lines, we propose a novel CDN paradigm utilizing lightweight Unikernel-based Virtual Machines (VMs), under which limited-content web servers hosted on Unikernel VM’s boot rapidly and on-demand, serve users’ requests and then shutdown. Although the first experimental results extracted from our relevant platform are promising, we have difficulties to conduct experimentation resembling real CDN deployments. FED4FIRE is the ideal experimentation environment to conduct such research, since it can provide all the aspects we are missing (i.e., scalability, heterogeneity of physical / virtual resources and low barrier of experimentation difficulty).
To summarize, the proposed work includes: (i) experimentation and a relevant demonstration of our unikernel-based CDN platform involving regular and lightweight clouds over two FED4FIRE+ test-beds (i.e., Virtual Wall and w-iLab.t) and the Utah Emulab test-bed in the USA, which is associated with the FED4FIRE+; (ii) implementation of two scenarios highlighting heterogeneity and scalability aspects, respectively; (iii) realization of a concrete data management plan that releases the produced data openly and enables further related research; and (iv) promotion of our research outcome and the FED4FIRE+ facilities through at least 3 scientific publications and demos in top research venues.
MATERIALS
- Real-Time Algorithms for the Detection of Changes in the Variance of Video Content Popularity (download the paper)
- UNIC experiment (download the poster)
- Review UNIC experiment | FEC6 (download the poster)