A few background comments spurred Faultline to go digging around in the forecasting archive of our sister service Rethink TV, regarding the actual cost of CDNs. More importantly, with the advent of WebRTC, Distributed CDNs (D-CDN), and some slightly more out-there approaches from the Web 3.0 crowd, it does rather look like the pressure is on for the CDN vendors.
However, they are not exactly in the dark about this. Akamai, the largest, has made no bones about its shift into the services realm, focused initially on cybersecurity but moving into the financial domain. A CDN is, after all, a distributed collection of servers, and it turns out that if the likes of Akamai add an infrastructure provider like Linode to the table, then a CDN can do just about anything it sets its mind to.
The other arrival that CDNs have to deal with is the popularity of live OTT content, which is just about outpacing on-demand video in their workloads. Complicating matters is the fracturing of audiences, with viewers splitting into smaller pools of people that are still expecting broadcast-quality viewing experiences.
Video from the non-broadcast world has massively improved in production values, over the past decade, and is increasingly poaching viewing time from the broadcast and Hollywood realm.
As such, CDN pricing is under pressure, and has unsurprisingly fallen on a per-volume basis. There are a number of different choices for CDN deployments. Public CDNs are multitenant, meaning that your video traffic is going to be sharing the virtual pipes with others.
Private CDN is single tenant, and includes both vendor-provided and in-house choices.
A Pure WebRTC approach relies on its namesake technology, and does not rely on dedicated CDN infrastructure, although it is providing much the same function.
In a similar vein, D-CDN is using a wealth of distributed and decentralized third-party hardware to create a content distribution network. A multi-CDN deployment does just that; it uses multiple choices or vendors to ensure optimum performance.
Typically, a public CDN is a good starting point for a video service, and often the user will eventually work their way up to commissioning a private CDN – once a scale has been reached that this makes economic sense.
Larger video services and operators like having control over these networks, but the smaller services do not have the scale to make such a move.
Rethink TV projections suggest that total video traffic will grow from 1,403 exabytes (EB) in 2020 to some 12,599 EB in 2027.
Within that total, public CDNs grow from 460 EB to 3,804 EB, and private CDNs grow from 926 EB to 7,027 EB.
The challengers to these include a pure WebRTC approach. which is projected to hold 1,519 EB by the end of the period, and D-CDN, which Rethink TV reckons will reach 249 EB.
As for revenues, these services combine to reach $42.5 billion in 2027, from around $8.5 billion in 2020.
Unsurprisingly, the ratios of revenue between the approaches follow the volume of traffic, but it is worth examining the trends here.
For public CDN, the revenue per EB declines steadily from $4.3 million to $2.9 million through the period.
Private CDN declines much more sharply, from $6.9 million to $3.8 million in 2027.
Pure WebRTC starts out at around the same level as private CDN, on $6.7 million, but ends up cheaper than both public and private CDN, as the scale of the technology builds – finishing at just under $2.9 million.
Finally, D-CDN starts at $3.8 million, and ends at $1.2 million.
When viewed collectively, the revenue per EB declines from just over $6 million in 2020 to around $3.4 million in 2027. Again, this is not surprising, but it might alarm those who have not pondered the issue before.
Because of the unrelenting growth of video consumption, the delivery networks are being forced to find more efficient ways to handle these workloads. CDN offload is a big one, but is best suited for on-demand content that can be cached. Encouraging the use of WebRTC for live streams would be beneficial, and is especially feasible now that ABR and DRM functions are beginning to be added by the vendors.
As such, the CDN market can still grow strongly, even as the effective price per exabyte falls. In addition, the CDNs can offload as much traffic as possible, to defer capital expenditure on infrastructure upgrades. Technologies like Open Caching might also open new efficiency and monetization doors, although there is still a good amount of dissent surrounding the feasibility of Open Caching in particular.