Fastly powers Layercake edge shift for live streaming

Fastly has deployed its edge cloud platform at Layercake, a Sydney-based digital engineering company that builds automation and orchestration tools for media production and distribution.

Layercake said the deployment has shifted some processing away from central infrastructure and out towards the network edge. The company linked that shift to performance and scalability during traffic spikes tied to live events.

Layercake sells a platform called Streamcake. The company positions it as a way for broadcasters, sports organisations and enterprises to manage workflows for live production and delivery. It said the platform automates parts of infrastructure deployment and distribution across multiple endpoints.

Live streaming workloads often show steep rises in demand at the start of a match or programme. Those peaks increase pressure on video delivery systems and related controls such as access checks. Layercake said it needed a model that handled those peaks without building large amounts of capacity at its origin systems.

"Using Streamcake, we deliver solutions for consumer applications where there is a huge spike in demand for video, especially live video," said Domenic Romeo, Founder, Layercake. "We see traffic for major events spike incredibly at the start of an event."

Edge deployment

Romeo said he first used Fastly technology while working on a project for an Australian sporting organisation. Layercake has since become a Fastly customer, and the company described the relationship as a basis for further development work at the edge.

"We're working with the Fastly team on innovative solutions at the edge, and that's extending our capability to provide customers with highly intelligent, highly dynamic solutions for stream play out at the edge," said Romeo.

Layercake said the edge deployment changed how it designs parts of its infrastructure. The company said it can move some processing and logic away from the origin and closer to users.

"Fastly has really added value to the solutions we're building where we can now confidently say that we can shift some of our core processing and intelligence from the origin, to the edge," said Romeo.

Romeo also linked that approach to operating costs and expansion plans. "It's really lowering our cost of operating the platform, as well as giving us the ability to grow at a pace that would be potentially inhibiting if we had to do this ourselves at the origin," said Romeo.

Vendor testing

Layercake said it ran performance tests before selecting a supplier for edge functions used by Streamcake. It said it benchmarked multiple vendors and compared response times and data access speeds for products described as Key Value Store and Object Store.

"We ran a number of benchmarks across a number of vendors and Fastly outperformed all of them," said Romeo.

Romeo highlighted object storage read performance in particular. He said that speed affected what the company could build and how it would behave under large surges in traffic.

"The performance of Fastly Object Store was astounding. We're talking about read times well under one millisecond, which allows us to build intelligent solutions using information at the edge that allows us to not only perform well, but to scale well when large traffic hits it," said Romeo.

Authentication focus

Layercake also said it is working on a streaming authentication approach using Fastly's edge compute. The company pointed to inconsistencies in how applications handle stream authentication, and it said that fragmented implementations complicate security practice across platforms.

"We're using Fastly's Edge within Streamcake to develop that, which allows multiple authentication methods behind the scenes to be obfuscated with a single stream authentication framework," said Romeo.

He said the approach depends on edge compute scale. "At scale, that can only happen with the level of compute and capacity that a vendor like Fastly can provide," said Romeo.

Layercake said it worked closely with Fastly engineering teams as it validated design choices for its edge work. It described the work as new in its use of edge infrastructure for this part of live streaming operations.

"What we are doing with Streamcake is fairly novel. It hasn't been done on the edge before, and there were certain directions with the solution that we needed to go with and validate," said Romeo. "The Fastly technical team provided us with incredible support to not only validate what we did, but optimise it before we launched it."

Layercake framed edge performance as a requirement for its product direction in live production environments, where viewer demand can rise sharply and without warning.

"Without the level of performance and scale at the edge that Fastly can provide us, it would be impossible for us to provide that level of innovation that we persistently strive for," said Romeo.