With consumption rates for online video growing at unprecedented levels, operator CDNs are heavily dependent on their servers to intelligently cache content at the network edge and deliver it close to the end customer. However, to realize the full potential from their networks, these CDNs must understand how to accurately plan for peak demand and precisely provision capacity.
To do so, operator CDNs must be able to gain a holistic view of traffic across all their edge servers — also called points of presence (PoPs) — not only to understand how the load on each is trending but also to make sure that each edge location is serving the intended geography. Routing excess requests from a particular area through a server in a remote location can lead to video stream interruptions and other delays and inconveniences which detract from the end user’s experience.
The stakes are high. Inexact capacity planning often leads to service interruptions, reduced quality of experience, and inefficient use of network resources. To accurately predict peak demand and properly plan capacity then, operator CDNs must understand three essential metrics:
- Peak bandwidth: the value in bytes per second of the highest 5-minute average bandwidth
- Peak concurrent video streams: the maximum value of average concurrent streams of all the 5-minute periods
- Peak hit rate: requests per second, which provides a good sense for CPU utilization on the edge servers
Understanding these three factors in combination requires multidimensional analytics and reporting that can help operators to drill down to determine the popularity of specific assets, understand the content consumption patterns of their digital media customers and quickly determine the bandwidth demands imposed on each edge location.