By Bill Ver Steeg, Distinguished Engineer, Cisco Systems
We are proud to put down in writing what we believe to be the first Reference Design Kit (RDK)-based deployment of IP video. Oh, and it is the first IPTV system running on IPv6. And one of the first based on a combination of premises based products and cloud-based services. And it all went from concept to turn-up in 50 days!
The deployment happened in Europe, but if you’re in Las Vegas this week at CES, we will also be demoing it for customers at The Wynn Hotel.
What was involved:
Our customer wanted to showcase its brand new IPv6 network by delivering a world-class IP video experience. An all-IPv6 IP video system had never been deployed before, so this was a non-trivial challenge. We chose to use the leading edge components in RDK in the IPv6 environment. Our challenge: they wanted it in and complete in 50 days, from project start to subscribers using it. To meet this challenge we turned to a combination of our new Videoscape Cloud Services SaaS offerings and premise based solutions.
Let’s talk about the toolkit that allowed us to deliver this customized solution in such a rapid timeframe. First and foremost, the delivery required all of the components to work in IPv6–only mode. It’s no great secret that Cisco is highly focused on IPv6 (understatement), and our RDK based systems are no exception. As our customers migrate from IPv4 to Ipv6, all of our video products are being widely deployed in mixed IPv4-IPv6 environments worldwide. As can be imagined, there were considerable production, testing and integration challenges with working in a pure IPv6 deployment.
We started by taking the customer’s existing IPv4 IPTV video sources, and converting them to IPv6. In order to drive the RDK-based CPE, we generated IPv6 flows for use in the new deployment, without interrupting the legacy IPv4 service. The public/private cloud architecture that we use for our video infrastructure allowed us to design and build the encode/ingest system in the cloud, using cloud-based services for the initial turn-up , and then move critical components into the customer’s facilities.
It was our ability to deploy a hybrid cloud model, which enabled us to meet the short timeframes and the performance requirements of the system. It alone gave us the critical time-to-market advantages of a cloud-based deployment model, along with the performance and reliability advantages of hosting the critical components in the customer’s data center (the former headend).
Back to the CPE:
We had to make some changes to the set-top box (STB) kernel, RDK and application code to support an IPv6-only delivery model. Thousands of those of IPTV STBs were needed, in a very short timeframe.
Some of the networking technologies that we used were based on the work being done in the IETF HomeNet working group. The primary goal of HomeNet is to enable multiple ISPs to provide service via an arbitrarily complex network inside a subscriber’s home. To this end, there is a lot of technology in the pipeline. Some of this technology (like the baseline IPv6 connectivity and the prefix coloring work) is applicable to an IPTV deployment from a single SP. We integrated these components into RDK for use in this IPv6 IPTV deployment.
The first CPE software change was in the kernel underneath RDK. We had to provide a robust Quality of Service (QoS) on the advanced IPv6-only network. As part of this effort, we wanted to validate a new QoS design based on IPv6 prefix coloring. In that design, the voice, video and data flows are mapped to their respective, prioritized network queues using specific IPv6 address ranges (this is also known as “prefix coloring.”) Specifically, we added support for prefix coloring to the Linux kernel in use on the STB.
Next, we extended the URLs supported in RDK and gStreamer to support IPv6, and verified that the video URLs mapped the right QOS functionality to RDK’s video flows. Then, the moment of truth: We mated the Cisco set-top box to the new residential gateway and the customer’s new FTTx network. We’d seen the IPv6 prefix coloring technology in the labs and in and in controlled demos, but this was the first use of prefix coloring to separate voice, video and data in live customer homes.
This RDK-based project was the first ever deployment of video over IPv6 outside North America. We found that the RDK layer was quite robust. Once we made some very well bounded changes to the software to support URLs that had IPv6 addresses, the actual deployment of the newly IPv6 compliant middleware layer went quite well.
Once we married RDK to the rest of the systems in our lab and the public cloud, the system components were delivered to the customer site — in 35 days. Next was final system integration and test, while we staged the STBs for deployment. In the end, we did have some long days in the lab – but we hit the 50 day deployment target!
We had some interesting design decisions around the use of Any Source Multicast and Source Specific Multicast, as well as the usual hectic scene that is integrating new residential gateway software with new STB software, on a new network design.
So that’s the story. We think it’s a story of state-of-the-art video delivery — IP video, using IPv6 and RDK and using a combination of data center and public cloud based components. We were glad, and are proud, to have been a part of it. The rapid system turn-up shows the power of the new tools that are coming into play in the digital video world.
We’ll be highlighting RDK’s role in this deployment, and discussing the technical details with customers at CES this week.