Modelling the Value of JITEX: Just-In-Time Transcoding at the Edge

In today's CDNs, edge servers are more than just reverse proxies. Computing resources can be reserved at the edge at a lower cost thanks to adoption of virtualized technologies and the availability of datacenters deep in 5G networks. CDN administrators can then evaluate whether to run processing-intensive tasks on or near the edge servers. Through edge processing, traditional CDN management limitations can be fixed by using computing resources. Adaptive Bitrate (ABR) Over-The-Top (OTT) streaming pose a new challenge for CDN management due the plurality of assets versions. Precisely, the CDN is required to offer dozens of videos per asset in order to accommodate a broad range of client capabilities, which reduces cache performance and increases backhaul costs. A solution to this problem is to receive a pivot video and transcode it on-the-fly into different formats by the edge processing solution. In this paper, we model the impact of this solution known as just-in-time-transcoding at the edge (JITEX). To do so, an exhaustive system model is designed allowing to define diverse sets of settings, mimicking real-world streaming scenarios. The results indicates that, the use of JITEX seems to be less cost-efficient when exchanged with cache memory budget. However, when exchanged with ingress traffic budget, specially when the down-link is naturally limited, the use of JITEX appears to be more promising.

Published

2022-10

Content type

Original Research

Keywords

CDN, Just-in-time transcoding, Edge processing, Cache management.

DOI

10.5594/M001990

ISBN

978-1-61482-963-8