European Parliament
The European Parliament works with the Council of the European Union to debate and pass legislation spanning everything from artificial intelligence and defence to deforestation and human rights. With 720 members from 27 member states, streaming its proceedings to citizens means supporting dozens of languages, high volumes of video sources, and a deployment footprint that could shift between on-premises and cloud infrastructure at any point.
G&L Systemhaus, a systems integrator specialising in video and audio streaming, was brought in to build that platform, with Norsk as a core technology component.
The Parliament’s requirements ruled out any off-the-shelf streaming product:
As Alexander Leschinsky, co-founder and CEO of G&L Systemhaus, put it: “We needed something that works both on-prem and in the cloud. At the beginning of the project, we did not know how much we would eventually solve on-prem, how much we would move to the cloud, which workloads we would move to the cloud.”
G&L built the platform on Norsk for several reasons that compounded across the project:
Custom hardware integration. Norsk’s extensibility let G&L integrate Deltacast SDI input cards — the highest-density SDI option available — a requirement the project couldn’t work without.
Hardware-accelerated transcoding. Rather than relying on CPU-based processing, G&L paired Norsk with NETINT hardware accelerator cards, avoiding the extra integration work that direct support would have required.
Reusable components. G&L had already used Norsk on prior projects, including work for WDR, and was able to carry components like its Playout Manager directly into the Parliament build.
One codebase, two environments. Norsk behaved consistently across on-premises and cloud, so G&L could write once and deploy wherever the workload ultimately landed — essential given the deployment split wasn’t fixed going in.
On the audio side, the 32-language requirement meant decoding and recombining audio across multiple redundant video feeds — a 20-channel system with 32 audio tracks each works out to 640 audio channels to transcode. The final architecture split the load across NETINT cards for video encoding, Ampere ARM CPUs (128 cores) for de-interlacing and audio transcoding, and Deltacast cards for input capture, with Kubernetes clusters in two Akamai Connected Cloud regions (Amsterdam and Frankfurt) handling content management, access control, and social publishing.
Within that architecture, Norsk handles the media processing: recording incoming signals into a seven-day ring buffer, encoding them into SRT streams for both transcoding and internal multicast, packaging into HLS for Akamai, re-muxing selected audio into RTMP for social platforms, and clipping live content into on-demand files.
“Norsk enabled us to deliver a complex and sophisticated media processing architecture in record time with a high level of system availability and resilience. Working with alternatives would have cost us both more time and money.”
— Alexander Leschinsky, Co-Founder and CEO, G&L Systemhaus
The Parliament project reflects G&L’s broader “atoms and molecules” philosophy: build well-defined, battle-tested components once, then combine them for each new customer instead of starting from scratch. Norsk is the media-processing layer underneath those components — which is why Alexander says G&L will “definitely continue to use it as a backbone” of its work going forward. For teams facing the same tension between deployment uncertainty, custom hardware, and tight timelines, it’s the same case Norsk makes everywhere else: the differentiating logic belongs in your hands, not locked inside a closed product.
Get in touch to talk through your requirements.