Multi-CDN Video Delivery: Three Tips for Optimizing the Streaming Experience

The world of online video is bigger than ever, with live streaming experiencing 99% YoY growth in 2020. And it’s not just due to the COVID-19-induced lockdowns, as time spent streaming grew by 58% in 2019 even before the pandemic hit. But this rapid growth comes with a caveat: with so many live and on-demand streaming services available to consumers, the tolerance for spotty performance or unavailable content is lower than ever. In a highly competitive OTT streaming market, end users have plenty of other options to choose from if their preferred provider is unable to give them a reliable, high-quality viewing experience.
Given this increased demand and low margin for error, an ever-growing number of OTT broadcasters and streaming providers are implementing multi-CDN strategies to improve the reliability and performance of their video content. This article will showcase the benefits of deploying a multi-CDN architecture, as well as some helpful tips for how to set one up.
Multi-CDN Benefits
Choosing to switch to a multi-CDN video delivery deployment can be a complicated process, but is generally driven by the following two goals:
Redundancy & Availability
Suffering a service outage and dealing with the negative PR, angry customers, and loss of revenue and brand reputation is something that no business wants to endure. Yet there is no such thing as a CDN that can guarantee 100% performance, which is why Lumen recommends having at least two CDNs in the delivery mix so that if your primary one fails, there is at least one backup (preferably more) to serve your video content in its place.
Performance Improvement
Many CDNs are only able to offer faster start-up times, improved throughput, and lower latency in certain geographic locations or on specific ISP networks due to limited infrastructure footprints. This can be a big problem depending on where your end users are located and what networks they’re accessing your video content from.
The best solution for this is to select a global CDN provider like Lumen that can offer high performance all around the world. And with multiple CDNs in your stack, you can prioritize certain providers in specific locations so that users are directed to the one best suited to serve content in that region.
Using multiple CDNs can also ensure that the best performing one can be used at the start or during a video session. CDN performance can vary depending on a variety of factors including DNS resolution, network bandwidth and congestion, server-side configuration issues, or a number of other factors. With a multi-CDN deployment, these issues can be mitigated by pointing the user request to a different CDN that provides more efficient content delivery at that specific time.
3 tips for implementing a multi-CDN video strategy
Pick Providers that Work Well Together
The dynamic nature of a multi-CDN architecture means that your configurations and switching strategies have to be able to fit with one another to avoid any unnecessary headaches when a switch is made. This means consistency of headers, security tokens (each CDN must be able to work with the video player to generate a unique token for the different URLs), and custom configurations.
The configuration work required to ensure seamless transition from one CDN to the next is something that your providers should be willing to work with you on as an extension of your own engineering team; declining to do so or charging prohibitively for configuration assistance should be a red flag when comparing different vendors.
You also want to make sure that the multiple CDNs in your stack are compatible across different advanced features such as origin shield, domain masking, object storage, and others. Multi-CDN environments have a lowest common denominator aspect to them, meaning if you ever have to switch over from a feature-rich provider to a more basic one, you’ll lose the benefits of that feature-rich CDN.
Architect a CDN Mix to Suit Your Needs
Deploying multiple CDNs for the reasons already covered above can certainly improve your end-user experience and operational efficiency, but adding too many providers to your architecture can reach a point of diminishing returns from both a provisioning and budgetary standpoint. Additionally, using too many CDNs on a content library with long-tail usage characteristics can decrease CDN cache efficiency, leading to degraded performance and potentially increased origin load and egress costs.
The best CDN mix is unique to each company and varies widely depending on use case and cost-benefit analysis. Do you primarily serve live or on-demand video? Where are your core users located? How much would a micro-outage or dip in performance affect revenue? How much would a 1-second improvement in start-up time benefit your business?
The answers to these and other questions will go a long way towards determining how many CDNs you need to deploy. Two CDNs may be enough for some regional broadcasters, while providers with more global audiences may need several more. Others focus choose to serve their live video from one CDN and their VOD from another. Streaming providers with premium content may deploy more redundancies or buy more capacity to deliver those streams since the viewers have a higher expectation of performance.
In determining your delivery needs and doing your cost-benefit analysis, remember as well that CDNs often provide lower rates in exchange for more volume. This means that if you start using more vendors on a regular basis, you might incur higher cost-per-bit-delivered from your other ones as the volume of traffic on them decreases.
It’s also important to remember that all CDNs in your stack should have at least 5-10% of your content cached on them at all times (this is known as keeping the cache warm) so that if there is a widespread outage that engages the emergency CDN as a primary source of content, you won’t jeopardize the health of your origin with a “thundering herd” of requests.
Implement a Switching Strategy that Fits Your Video Services
The most important aspect of your multi-CDN strategy are the policies that you put in place to determine how the traffic is balanced between your different providers. This policy will ultimately be determined by budgetary considerations, quality expectations from your end users, and how your operations team is constructed (i.e. how much time and effort can be committed to managing the switching decisions).
There are three main criteria to base your CDN switching decisions on:
- Static business rules
- Commit-based rules
- Performance-based rules
Static business rules are the easiest to manage as they involve little to no operational effort after they’re set up. These are typically done based on specific types of streams, geographies, ISPs, or devices. They could also be set up for specific times of day as a form of budget management; for example, you can choose to use your cheapest CDN during non-peak viewing hours because you’re not as worried about performance issues during those times.
Commit-based rules are set up according to how much volume has been delivered by each CDN within a set time period (usually monthly) and how much is yet to be delivered to fulfill contractual commitments with that provider – as well as how much can be delivered before overage charges kick in. These switching criteria are usually managed via API, which means that the different CDNs in your stack must all be able to support that switching mechanism. In absence of an API, you can make the switches manually on the backend or institute hard thresholds that cannot be surpassed.
Making your multi-CDN switching decisions based on performance thresholds requires integration of additional software solutions that monitor the quality of service and can trigger a switch if one CDN vendor suffers an outage or a drop in performance. This QoS data can be collected with a third-party monitoring tool or a custom in-house solution that tracks real-time CDN metrics like time to live, round trip time, latency and errors, or player metrics such as buffering, startup time, track switches, or bitrate.
These different criteria should be weighed depending on the type of video being delivered as well as your unique business needs and processes. For example, live audiences tend to be more sensitive to degradations in QoS than video on-demand, so if that’s your primary type of content, then an ideal decision apparatus is one that leverages performance data in conjunction with simple business rules. Below is an example of a decision tree that could serve as the basis for a multi-CDN switching strategy:
An ideal switching mechanism is one that’s able to take multiple factors into consideration. A solution like Lumen® CDN Load Balancer does this by collecting real-time QoS metrics from the devices watching the stream and weighing them against business inputs on the backend. If the default CDN’s score drops to certain point, CDN Load Balancer will switch to the second option. Moreover, it does this mid-stream during the video session, thus eliminating the need for the viewer to reload the page and preserving the overall end-user experience.
The decision to switch to a multi-CDN architecture is one based on industry best practices and the need to provide the best end-user experience for your viewers, but that doesn’t mean that it’s a simple process. To do so properly, you must determine what goals you want to achieve with it and select CDN providers that offer the best coverage, the fastest performance, and willingness to as an extension of your team to help set up the best architecture to suit your specific delivery needs.
For more information on multi-CDN strategies and how Lumen can help enable them, reach out to content-delivery@lumen.com.
This content is provided for informational purposes only and may require additional research and substantiation by the end user. In addition, the information is provided “as is” without any warranty or condition of any kind, either express or implied. Use of this information is at the end user’s own risk. Lumen does not warrant that the information will meet the end user’s requirements or that the implementation or usage of this information will result in the desired outcome of the end user. All third-party company and product or service names referenced in this article are for identification purposes only and do not imply endorsement or affiliation with Lumen. This document represents Lumen products and offerings as of the date of issue.