Why CDN reliability is a business question
Streaming is real time. Unlike a website where a ten-minute outage means a visitor tries again later, a CDN outage during a premiere or live event means viewers close the app - and do not always return. Lost viewing minutes cannot be recovered after the fact.
This is why multi-CDN strategies are no longer the exclusive domain of the very largest streaming operators. Mid-sized and specialised platforms - sports channels, niche VOD services, educational streaming - increasingly recognise that the risk of relying on a single CDN is not proportionate to the cost of distributing traffic.
This article is a technical walkthrough for those responsible for architecture and operational reliability. We cover the risk profile of single-CDN, the two principal multi-CDN configurations, what is specific to Nordic streaming, how traffic steering and failover work in practice, and how to keep costs under control in a distributed CDN system.
Shapp works with streaming platforms and has previously covered fundamental infrastructure decisions in our guide to building streaming platforms.
Why single-CDN is a risk
A single CDN carries three categories of risk that a multi-CDN strategy addresses.
Infrastructure failures and service disruptions
CDN providers have outages. This is not a hypothetical risk - every major provider has had documented disruptions that affected significant portions of the internet. The question is not whether it will happen, but what the consequences are when it does for your specific platform.
For a platform broadcasting live sport or operating a paid tier with SLA commitments, the answer is straightforward: the consequences are unacceptable. For a pure VOD offering with a free tier, tolerance may be higher, but the brand impact of perceived reliability problems is always negative.
Geographic performance and PoP coverage
No CDN has optimal Point of Presence coverage in every region. A provider with strong coverage in the Stockholm area may have weaker performance in northern Finland or Iceland. In the Nordics - with a relatively sparse population spread across a large geography - PoP coverage is a genuine differentiator between CDN providers.
Single-CDN means accepting the geographic performance gaps of that specific provider. Multi-CDN makes it possible to steer traffic from specific regions to the provider that performs best there.
Negotiating position and pricing risk
A sole-provider relationship offers limited negotiating leverage at contract renewal. With multiple active CDN providers, you have real data on the alternatives' performance and cost - and a genuine option to shift traffic if pricing changes.
Active-active vs active-passive
There are two fundamentally different ways to organise a multi-CDN environment, and the choice between them depends on traffic volume, complexity tolerance, and economics.
Active-passive: the simpler intermediate step
In an active-passive configuration, a primary CDN handles all traffic under normal operating conditions. The secondary CDN is kept "warm" - that is, configured and tested - but does not carry production traffic until a failure is detected at the primary.
Advantages: Simpler to configure and monitor. Lower ongoing costs because the secondary CDN generates minimal traffic. Easier to troubleshoot because normal operating mode is single-channel.
Disadvantages: Failover time is longer. The secondary CDN is not continuously performance-tested with production traffic, meaning its actual performance under stress is difficult to predict. It does not provide the geographic performance advantages of actively selecting the best CDN per region.
Active-active: full distribution
In an active-active configuration, traffic is continuously distributed across multiple CDNs. The distribution can be static (for example, 60/40 per provider) or dynamic (driven by real-time performance and cost per CDN per region).
Advantages: Maximum redundancy - a CDN failure results in immediate rebalancing to the remaining providers without perceptible interruption. Real-time geographic optimisation. Provides continuous, comparable performance and cost data per provider.
Disadvantages: Requires a traffic steering layer (an intelligent load balancer or traffic management platform) that is itself a component to operate. Harder to debug performance issues when they are not reproducible on a specific CDN. Higher baseline costs.
CDN selection criteria for Nordic streaming
Choosing CDN providers for a Nordic streaming platform means weighing a number of factors that are partly specific to the region.
PoP coverage across Scandinavia and the Baltics
Verify actual PoP placement in Stockholm, Oslo, Helsinki, Copenhagen, and if relevant Tallinn and Riga. More importantly: verify peering arrangements with the dominant Nordic ISPs. A PoP in Stockholm that lacks good peering with Telia, Telenor, and Elisa will deliver worse performance than the PoP list suggests.
Support for adaptive bitrate and CMAF
Modern streaming infrastructure is built on CMAF (Common Media Application Format) and HLS/DASH with adaptive bitrate. Verify that the CDN provider supports CMAF with low-latency chunked transfer, has configurable cache rules per content type and segment length, and handles origin shielding correctly to minimise origin load under high concurrency.
SLA and support response
SLA guarantees for uptime and latency are one thing. Actual support response during an incident is another. Ask for reference customer contacts and put specific questions about incident response times and escalation paths. A 99.9 percent SLA without a clear incident management process is a piece of paper.
Pricing model and egress costs
CDN pricing is complex and varies significantly between providers. Egress cost per GB is the most obvious metric, but cache hit rate, minimum volume commitments, and pricing tiers at high volume affect the actual cost equally. Request a cost comparison based on your actual traffic profile - the mix of VOD versus live, average bitrate, geographic distribution.
Traffic steering and failover
How traffic is distributed and redirected is the core of a functioning multi-CDN strategy.
DNS-based steering
The most common method is DNS-based traffic steering via anycast or geolocation DNS. An intelligent DNS resolver returns the CDN endpoint that best matches the viewer's location and current CDN health. Failover occurs when health checks against a CDN endpoint begin failing and DNS responses are updated accordingly.
TTL (Time to Live) for DNS responses is a critical variable. Low TTL (30 to 60 seconds) enables fast failover but increases DNS query load and can reduce cache hit rate for individual sessions. High TTL reduces failover speed.
Token-based manifest manipulation
A more sophisticated approach is to steer CDN selection at the manifest level. The player fetches a master manifest from an origin server which, based on real-time health data and traffic steering logic, returns segment URLs pointing to the most appropriate CDN for that session. In the event of a CDN problem, the manifest is updated with alternative URLs at the next manifest request.
This approach provides finer control and is more transparent to the viewer, but requires more sophisticated backend logic.
Health checks and alerting
Automated health checks against CDN endpoints should be configured with appropriate frequency (every 15 to 30 seconds for live, every 60 seconds for VOD), verify actual video delivery rather than just an HTTP 200 response on a test file, and trigger alert flows with sufficient lead time to allow manual intervention if automatic failover does not activate as expected.
Cost optimisation in a multi-CDN system
Multi-CDN does not have to mean double the cost. With the right strategy, total CDN cost can be kept equal to or lower than a well-negotiated single-CDN contract.
Volume commitments and blended pricing
Distribute traffic in a way that maximises volume discounts with each provider. If provider A offers a discount at 500 TB per month and you normally deliver 400 TB, it may be worth steering more traffic there to reach the threshold.
Content classification and caching strategy
Popular content (the top 10 percent of the catalogue) generates 70 to 80 percent of traffic. Ensure that this content has an optimal cache hit rate across all CDNs. Long-tail content with low demand can advantageously be served via a single, cost-efficient CDN without risking performance impact on the broader viewer experience.
Origin shielding
Origin shielding - an intermediate layer that aggregates CDN requests toward the origin - dramatically reduces origin load and bandwidth costs for popular content. In a multi-CDN environment, origin shielding is configured separately per CDN but can be coordinated against a shared, cost-optimised origin.
Summary: CDN as strategic infrastructure
Single-CDN is a reasonable starting point, but for a streaming platform with ambitions it is a risk that grows with the platform's success. The more paying viewers, the more expensive every minute of unavailability becomes.
Multi-CDN strategy is not a single decision but a system of choices: configuration model, provider selection, traffic steering logic, and cost optimisation. Each part affects the others, and the system needs to be designed as a whole.
This is precisely the type of architecture Shapp helps streaming platforms design and implement. Contact us to discuss what the right CDN strategy means for your specific situation, or read our guide to building streaming platforms for a broader overview of infrastructure decisions.
Get in touch for a no-obligation conversation.
