Name
Modelling the value of JITEX: Just-in-time-transcoding at the Edge
Date & Time
Thursday, October 27, 2022, 2:30 PM - 3:00 PM
Gwendal Simon
Description
The edge servers in today’s CDNs are no longer basic reverse proxies. The adoption of virtualized technologies and the availability of data-centers deep in 5G networks makes it possible to reserve computing resources at the edge, for certain time periods, at a lower cost. CDN administrators can then evaluate whether or not to run processing-intensive tasks in or near the edge servers. The idea behind edge processing is to exploit these computing resources to fix the limitations of traditional cache-based CDN management.
One common problem that occurs in CDN management, when it comes to adaptive bit-rate (ABR) OTT streaming, is the number of varying versions of a singular video asset. To meet the needs of a wide range of client capabilities, the CDN has to offer a dozen videos per asset, which reduces cache performance and increases the burden on the CDN backhaul.
To solve this, academic researchers have proposed an edge processing solution where the edge receives only one pivot video and transcodes it on request to different formats.
In this paper, we will model the impact of this just-in-time-transcoding at the edge (JITEX).
After a technical overview, we will identify the cost variables that are associated with the deployment of transcoders in network edges, along with CDN costs. For transparency, we will only use public numbers and anticipate their variations. With this model, we then identify the use cases for which JITEX is valuable and describe the most attractive conditions, both from the research and market perspectives.
By sharing this model, we want to further the debate on the viability of deploying transcoders at the edge. With a scientifically neutral standpoint, we will debunk the myths behind edge processing to highlight the opportunities, and identify the key research themes that impact the profitability of transcoders at the edge.
Location Name
Salon 1
Take-Aways from this Presentation
Attendees will: • Learn the benefits, opportunities and impact of just-in-time-transcoding at the edge • Evaluate cost variables associated with the deployment of transcoders in network edges and the CDN costs • Identify key research themes that impact the profitability of transcoders at the edge • Debunk the myths behind edge processing