Stef van der Ziel, founder and owner of Dutch content delivery network (CDN) services provider Jet-Stream, on the lack of profitability for Internet CDNs from media delivery, the (lack of) business case for CDN federation, and the pitfalls of licensed CDNs.
What is a broadcast-grade CDN?
A broadcast-grade CDN is a CDN that: runs on a telco network instead of the best effort Internet; uses active managed, guaranteed distribution technologies instead of best effort caching; and uses active managed, guaranteed request routing technologies instead of best effort DNS.
This guarantees content providers and subscribers availability, performance and capacity similar to digital cable / IPTV services. Typically these CDNs offer 99.999% uptime service-level agreements (SLAs).
We see a trend where Internet CDNs stop investing in media delivery (reshifting focus into clouds, app acceleration, enterprise) and telco CDNs focus on regional premium media delivery, both for their retail services and for wholesaling their CDN to OTT providers.
Internet CDNs have been unable to become profitable from media delivery. Telcos have a serious opportunity here, but just extending an Internet CDN into their network or deploying Internet CDN technology on their network is not enough. Subscribers demand QoS, and don’t like QoE: QoE is a euphemism.
We also predict that these premium CDNs will deeper interact with the underlying IP network to dynamically exchange QoS information on a per session basis and will also deeper interact with clients to instruct them to maximize uptime and performance dynamically.
How do you think patching issues between CDNs, carriers and access providers can be resolved?
Realistically they will not be resolved in a serious way. The business case for CDN federation is very weak. Content is actually extremely local due to rights, languages and culture.
Some vendors try to load balance over multiple Internet CDNs but that is not true federation. Telcos try to come up with standards via IETF and ETSI to interconnect their CDNs but these standards are based upon the lowest common denominator = poor.
Internet CDNs want to deploy edge nodes within telco networks but telcos do not allow it. Telcos also start charging more for inbound traffic and hosted servers instead of offering free peering.
How can CDN capacity be most effectively introduced into a telco’s network?
Internet CDNs will have a hard time convincing telcos, which have the choice to implement their own CDN.
What we have seen is that it doesn’t work when a telco starts R&D for a CDN from scratch. It costs them millions, years of time to market and they are not getting a better CDN than what Internet CDNs can offer. What also doesn’t work is to partner with an existing Internet CDN.
Some CDNs have a service which they call a ‘managed’ or ‘licensed’ CDN – however in reality the telco doesn’t host, own and operate the control plane, it is just an extension of the CDN’s edge nodes into the telco CDN.
Telcos thought this was an easy entry into the market but in reality they are not making any money out of it, lose their customers to their CDN partner, and the CDN partner gets a free ride on the network.
We also have seen telcos work with traditional system integrators and vendors to deploy a CDN, but that has failed because these companies don’t understand the extreme dynamics of a CDN as opposed to how static IP networks and IPTV platforms are.
What does work is to start with strategy and business consulting, so on C-level the telco gets knowledge about how a CDN can help reduce traffic load and monetize the network at the same time and how the telco can identify a good strategy.
Then we deploy a proof of concept so the telco gets some operational experience. Then we either deploy a full CDN in their network and manage it for them; or we train their staff so they can operate the CDN on their own.
So it is actually not so much about the network, or the CDN technologies, but about the knowledge gap. Making the right choices, identifying the pitfalls and going in with a clear strategy.
What effect does this have on service-level agreements?
Even when Internet CDNs deploy servers into telco networks, they still rely on best effort technologies such as DNS and caching, so placing edge nodes into telco networks does not improve their uptime and performance. It’s like putting a horse carriage on the German Autobahn.