Jeff Webb

Work

I've spent much of my career designing how live TV reaches people's screens. This page covers the streaming platforms I worked on at Sky, the decisions behind them, how I've shared what I learned and where it all started.

  1. In-house video-delivery platformEarly streaming ran on vendor appliances in the data centre, and as more people started watching, it did not scale well enough. Between 2010 and 2012 I designed the replacement: a dedicated video network (the DVN), an in-house video-delivery architecture. Its capacity was organised so it could grow and so a fault in one area stayed in that area. I partnered with the network and engineering teams who built and ran it. Sky Go and NOW TV ran on that foundation.
  2. Portable, software-defined streaming platformWhich cloud provider to use was a business decision, and it changed when the business changed. I led the solution architecture for the software-defined streaming platform, where each channel runs as software on shared servers rather than on its own dedicated equipment. I designed it to be portable, so it could move without being rebuilt. I partnered with Site Reliability Engineering (the team who keep live services running), network design and the engineering and operations teams who implemented it. I co-presented it running in Google Cloud in 2019. I designed it so a later cloud decision would not mean starting again. The trade-off was real development and testing effort, with partners who were still early with microservices and Kubernetes (ways of splitting software into small parts and running them on shared servers). In return, the platform could follow the business without rebuilding every channel.See the talk: Google Cloud Next ’19 co-presentation
  3. Single-channel fault domainOn older set-ups, one failure could take several channels off air at once, and for live TV that was too much to lose. We called the highly available design the single-channel fault domain: a synchronised live channel across two data centres, with the system able to self-heal without affecting customer viewing. The same live channel ran in two places at once, with both copies kept in step, so viewers could be switched to the healthy copy if one had a problem. It needed more testing, but it improved uptime. It also let the Site Reliability Engineering team move channels across during software updates, so an update did not have to take a channel off air.
  4. Teaching and onboardingPeople from lots of different parts of the business kept asking me the same question: how does streaming actually work? Technology teams quickly build their own vocabulary and tend to assume everyone else already understands it, but in the real world people need time to learn it. It's a bit like me speaking French to someone who speaks English: unless they learn French or I translate, we can't really communicate. We were a TV company, but we were also among the early movers in streaming, so I saw a real need to translate. In 2017 I worked on an animation to get the main ideas across, and in 2018 those ideas grew into a five-part course on how live streaming works, from the camera to the screen. It follows a live picture step by step and explains each part in plain words as it comes up. The aim was simple: to demystify the inner workings of a streaming platform at a very large broadcaster.
  5. Conference speakingSelected speaking includes a co-presentation at Google Cloud Next ’19, the Streaming Forum 2018 opening keynote, a TV Connect 2017 panel and a talk on video quality at the Sky Technology Conference 2017.

Before Sky

I started out in 1988 at AppleCentre Preston. In 1995 I co-founded UK Internet Provider Ltd with friends and was its Technical Director until 2004, looking after web hosting, broadband, email and tech support. Along the way I taught myself networking and servers. That experience led me into network engineering and then to streaming at Sky.

There's more of that story on the About page.