Three Key Strategies for Streaming at Scale 

John Agger

Livestreaming has become a core component of the broadcasting ecosystem in recent years. Almost a quarter of global online viewing time consists of live content, with the industry expected to continue its rapid growth throughout the rest of this decade.

For the companies whose business is delivering livestreamed content, ensuring their service runs smoothly is the number one priority. Building and maintaining a high-quality service is an iterative process that begins well before hitting ‘go-live.’  This process requires regular assessments and updates by the tech teams behind the scenes. But how exactly can they ensure that their biggest live streams are secure, reliable and are providing viewers with the best possible experience?

Dig into the data

You only get one chance when broadcasting a livestream. If traffic is higher than usual and you haven’t prepared, there can be serious consequences: your audience could experience buffering, drops in video quality or even a full-scale outage. Working with vendors that can give you important visibility into critical metrics will give you the best odds to react and resolve issues quickly.

With a thorough, data-driven approach you can choose where to focus your attention and accurately forecast peaks in activity. For instance, recording expected peaks in viewer numbers, engagement rates, and most common viewer locations during high-profile live TV shows allows you to form ‘best case’ and ‘most likely’ estimates for strategising future efforts.

With experience, it becomes easier to choose which metrics to monitor and prioritise. Industry experts can also offer advice to help you select which data to leverage for future events. This can help you choose where to focus, allowing you to make data-driven decisions, such as time-to-first-byte, error counts or latency, among any number of other variables. By communicating effectively and using historic data to identify and resolve issues, you will be better able to safeguard against future reappearance and ensure continued delivery of top-quality streaming experiences.

Eliminate single points of failure

Failure in isolated parts of the network can cause the whole stack to fail, resulting in a poorer quality stream, potential outages and ultimately lost revenue or reputational damage. An infrastructure prone to failure can go unnoticed, which will cause serious problems if your system lacks redundancies. CDNs delivering your stream around the world are vital to efficiently scaling up when more viewers join. If these contain single points of failure, your stack might not be able to efficiently distribute your video files over the available bandwidth, meaning you lose the scalability that CDNs provide and the user experience suffers.

Adding redundancies to your livestreaming arsenal by working with multiple delivery networks allows you to scale up quickly without fear of failures once traffic starts building. This larger architecture means you can easily switch to a backup network and also lets you split or reroute traffic around congestion points.

Bear in mind that this approach should be applied across your full stack – including capture devices, encoders and players – not just the point of delivery. Incorporating as many redundancies as close to the origin server as possible will give you a stronger, more resilient network to run your livestreams on.

Pre-test to de-stress

Being disciplined about testing and safeguarding the video delivery stack is crucial to preparing for smooth online delivery. This includes load testing and implementing configuration changes on-the-fly for everyone involved, including technical partners.

Smaller-scale events, like off-season or regional sports matches, provide a perfect opportunity to assess your workflow, configurations and processes. These dry runs should include all components of your technology and services, from checking your delivery pipeline, to logging (both for troubleshooting and for post-event measurements) and connectivity.

This time can also be used to ensure operational contingency plans work if needed. They need to be documented, vetted, disseminated and practised more than once in order to be utilised effectively. You should be thinking about what you will do when action is required. If you need to reallocate traffic between CDNs, how will this be done? How will you monitor real-time data, quickly identify issues and determine location? Real-time messaging channels with third-parties can again be useful here if a test results in a decision needing to be made.

Deliver consistently

By using these strategies there is a lot you can do before, during and after a live event to help ensure a successful outcome. Every livestream presents an opportunity to collect performance data, look for faults in the network, and test the delivery stack. Everything learned from carrying out these processes feeds into how the next stream is run, and the result will be an infrastructure that can expand safely and dependably when needed with a solid backup plan for any unwanted surprises.

This article was written by John Agger, Fastly, principal industry marketing manager for media & entertainment.