As online video viewing reaches new heights, the challenge for broadcasters grows. Live events continue to set streaming records, while new video on-demand platforms are hitting app stores faster than consumers can keep count.
To keep up with growing demand throughout the world, an increasing number of video providers are turning to multi-CDN architectures. Relatively little, however, has been documented on the subject. Eager to know more about where the industry is heading, we recently sat down with David Hassoun, Founder & CEO at RealEyes Media. An expert in the field, David has decades of experience in online video: developing digital media applications, teaching programming at the University of Denver, and founding RealEyes Media over 15 years ago. Today, his Colorado-based company advises some of the US’s largest media groups on complex technology implementations.
The following interview details David’s experience with multi-CDN.
Streamroot: how is multi-CDN perceived by the largest broadcasters?
David: For our top-tier customers, multi-CDN is not a question nor an option, it’s a necessity. Multi-CDN has either been in play, or they are moving to make it part of their standards immediately. They are either opting for an in-house solution to manage everything themselves, a hybrid solution leveraging data from third-party vendor APIs, or full out-of-the-box that they can just put in place and have it work.
Do they tend to develop their solutions in-house or via third-party vendors?
I would say 50/50 right now. VOD tends to be more off-the-shelf than live.
Do you see mostly static or dynamic switching strategies?
Everyone wants dynamic but they often struggle to reconcile the complexities involved. I see static as a gap-fill for those planning to do dynamic. For anything off-the-shelf, broadcasters generally require dynamic since it is where they want to go anyway.
In your experience, are broadcasters mostly motivated by cost, quality, or a combination of both?
Both is the real answer, but in my experience it has been more about quality. Cost is a factor, but less and less as CDNs have become commoditized and costs have dropped.
The subtext on this, however, is that quality shouldn’t always be an absolute goal. Sometimes broadcasters focus too heavily on this, wanting to go to the CDN that provides the absolute optimal performance at an exact point in time. Switching too aggressively often ends up causing more problems than it solves, and the incremental improvement is so low that the return isn’t there. The target that we usually recommend is consistent experience.
Phase one of a multi-CDN approach is therefore having a full multi-CDN failover and a seamless route, with some logic for selecting your CDN upfront. Because these problems tend to come and go in relatively short periods of time, if you are maintaining your standard quality of service… stick with it! You are going to have peaks and valleys, and that is normal. If there is an actual issue, react. Otherwise don’t worry about it. We often find that the tools out there switch too aggressively whereas the customer may simply want a multi-CDN solution in the event there is a problem, especially on the live side.
In media and entertainment [VOD], on the other hand, customers will often want to take this a step further as they operate in a very competitive landscape where quality can directly affect viewership and subscriptions. With all of these new platforms popping up, they have to be optimal; if they have an easier route to do that, they can harness certain benefits. For live, it is more “Go for it, don’t mess it up, and if there’s a problem have a back-up ready and rolling.” vs. for media and entertainment, it’s more a question of “What’s the optimal route that we can get to shave off a hundred milliseconds?”
Are there any perceived technical barriers to implementing multi-CDN? On the flip side, do you find that broadcasters are cognisant of the effort it takes to deploy a multi-CDN strategy?
The bigger ones understand how complex this is, whereas the smaller ones tend to opt for turnkey. The question is more about perceived vs. real complexity. Some “turnkey” solutions tend to oversell the simplicity of implementation whereas the reality is never that simple under the hood, especially when dealing with the rolling impact it can have across other systems.
For example, you’ll need to have the same token system across all CDNs, which is rarely the case when you start out. Or some platforms may want to roll out multi-CDN on a specific product or platform, but even then, these changes impact a lot of different teams across the organization.
Many start out thinking it’s easy but then start considering what they are trying to accomplish and how that actually breaks down. It means determining the business logic and what will drive the switching, considering to what extent the solution will be deployable across platforms, and potentially turning over the keys to their DNS to a third party, etc.
One of the biggest challenges is that broadcasters need a solution that works across all platforms. Some time ago smartTV compatibility wasn’t an issue… I miss those days! Another challenge is that they don’t want to have multiple integrations, but rather something all-encompassing in some way or another. Some customers say, “As long as we have iOS, tvOS, Android, Fire TV and web, we’re good.” At the bare minimum, those primary platforms are key.
What are some common pitfalls that you see?
Token security is one thing that is sometimes overlooked. And ad workflows: have you figured out how implementing multi-CDN affects your ad calls? Especially in CSAI [Client-Side Ad Insertion] there is a huge layer of complexity there. But even in SSAI [Server-Side Ad Insertion] there is a challenge of how to do the bootstrapping and handle multiple CDNs, and SSAI providers have not necessarily worked through all of the use cases yet. How you deal with sessions being killed, hopping over when it’s basically a unified session, balancing ad loads across different services, DVR… there are a bunch of things that are easily overlooked.
What is the minimum threshold for multi-CDN to be a necessity?
I think the baseline is anyone who wants to monetize an event and make sure that is has better resilience by not putting all of their eggs in one basket. The cost of making the platform more resilient versus return on investment is another question to ask. For me, the entry level is resiliency, the scale level is cost management.
Do you think that the current solutions on the market sufficiently address the different multi-CDN use cases?
With limited experience in off-the-shelf technologies and more popping up nearly every day, I don’t think one solution offers all the different types of options you want. From what I see, there is still a lot of ambiguity around the decisioning in the business layer, how that’s driven and with what control. The other area where I see tremendous opportunity is speed of which an issue can be identified and users moved. There is a lot of flux around the time it takes from issue to resolve, and most solutions are in the minutes which can feel like forever and may be too slow in live.
Will the trend towards multi-CDN architectures speed up in the coming years?
There is going to be a lot more movement towards multi-CDN architectures, and we are already seeing new players pop up. I do think this trend is going to speed up and that it is going to be a crowded market for a while. It will be interesting to see how platforms adapt to have more of the capabilities needed.
Do you have anything else to add on the subject of multi-CDN architectures in the video industry?
It’s a necessity and a challenge and it hasn’t been standardized yet!
As viewership grows and users demand higher and higher quality, broadcasters of every size need increasingly sophisticated multi-CDN solutions. A big thanks to David for sharing his insights on this growing market and important trend. Stay tuned for more multi-CDN research coming up this Fall.
Interested in contributing your own insights to our content? We are currently conducting a poll of broadcasters about their video delivery workflow. Share you experience in our anonymous survey here and sign up below to be the first to receive our report that will analyze the multi-CDN strategies used by streaming platforms across the world.
And if you are using or planning to implement a multi-CDN setup, consider trying out CDN Orchestrator (formerly Streamroot Compass). Our multi-CDN selector allows for instant and seamless switching within the stream based on real QoS feedback from the device!