Streamroot uses Google Kubernetes Engine and Google Cloud Load Balancing to handle peaks in traffic for its global peer-accelerated content delivery solution that powers over 20 million video sessions a day with zero downtime.
Google Cloud Results
- Autoscales using Google Kubernetes Engine to manage immediate peaks in demand across time zones
- Manages global traffic using Google Cloud Load Balancing to optimise video delivery and eliminate downtime
- Improves time to market by using integrated Google Cloud Platform tools such as Google Stackdriver
Handles over 40,000 http requests a second with Google Kubernetes Engine
Offering an innovative peer-to-peer solution to deliver high-quality video over the internet, Streamroot is now the service of choice for a number of major broadcasters to ensure the smooth streaming of large events. It currently powers 20 million video sessions a day and is continuing to grow with customers on every continent. Founded in 2014 by three French engineers, Streamroot has offices in Paris, New York, and Denver.
Streamroot works alongside traditional video delivery that uses servers and content delivery networks (CDNs), but instead of a one-to-many architecture it distributes the load amongst viewers by enabling more efficient peer-to-peer exchanges via WebRTC. This significantly offloads CDN traffic, resulting in a lower cost for the broadcaster as well as better quality video for the viewer, and potentially infinite scale, with a robust delivery network that grows stronger as more viewers tune in.
“We’ve experienced significant growth since early 2017, which translates to a massive increase in our traffic. We needed an architecture that could scale very quickly to this kind of traffic, which is global and peaks across time zones. Kubernetes was the answer.”—Reda Benzair, VP of Engineering, Streamroot
Streamroot’s customers are major broadcasters, such as DailyMotion and Canal+, that expect flawless video delivery for major events such as season finales, the Superbowl, or this summer’s highly anticipated World Cup, which Streamroot is due to power for some of its tier-one customers. Since Streamroot delivers its services globally, it needed a way to autoscale in response to peaks in demand and to distribute its load across a global network. Google Cloud Platform was the perfect fit.
“We have grown exponentially since the beginning of 2017,” says Reda Benzair, VP of Engineering at Streamroot. “We needed an architecture that could scale instantly to this kind of traffic, which is global and peaks across time zones. Kubernetes was the answer.”
Building a global architecture
When millions of viewers are watching a live sports event, broadcasters need their video streams to work flawlessly with no buffering. By using proprietary matching algorithms Streamroot technology intelligently leverages the user internet service provider, as well as network topology, device, and bitrate to determine the fastest and most efficient connections. Content multisourcing promotes a more efficient use of the network, safeguarding against server saturation and providing viewers with a better quality of experience. As a result, viewers enjoy up to three times less rebuffering than with traditional CDN-only delivery.
To deliver its services globally to millions of viewers, Streamroot looked for a robust architecture to autoscale immediately and handle peaks in demand. Its previous cloud provider struggled to cope with its rapid growth in users. “We quickly reached a point where we needed to rethink our backend architecture in order to scale up and handle that kind of traffic,” explains Reda. “We needed to manage the growth in traffic in Asia, the United States, and Latin America with peaks during the night and in the morning.”
“Auto-scaling absorbs any growth in traffic very efficiently, and when traffic decreases Google Kubernetes Engine shuts down smoothly. There is no money wasted for these services.”—Reda Benzair, VP of Engineering, Streamroot
To do that, Streamroot switched its core services to a microservice architecture on Google Cloud Platform using Google Kubernetes Engine. Thanks to Google Kubernetes Engine, Streamroot can autoscale at any time to meet peaks in demand and with Google Cloud Load Balancing, it can handle instantaneous traffic with low latencies. “We literally have viewers in every country of the world—we’re delivering all over the globe,” says Erica Beavers, Head of Marketing and Partnerships at Streamroot. “The ability to scale regionally is extremely important to us.”
Before, Streamroot had to spend time building telemetry tools around its architecture. Now, it uses Google Stackdriver for integrated logs aggregation and monitoring. “Having all these tools fully integrated with Google keeps the time to market short for us,” says Reda. “We didn’t have to spend time building a new logging and monitoring platform. Instead we spent time learning more about it, configuring it and using more advanced settings.”
Autoscaling also means Streamroot only pays for the services it uses. “Auto-scaling absorbs any growth in traffic very efficiently and when the traffic decreases, Google Kubernetes Engine shuts down smoothly,” says Reda. “There is no money wasted for these services.”
Easy integration and support
In addition to these technical advantages, Google Cloud Platform offers Streamroot operational benefits. “Google Kubernetes Engine brings a lot of value for DevOps. It is very easy to use, the documentation is well-structured, and the engineers feel very comfortable working with it,” says Reda. “Google is always up-to-date with the latest versions, which is important for our operations, and Google Cloud Platform offers smooth integration with many other useful products.”
Streamroot also appreciates the technical expertise and support Google offers. “We spent a lot of time working with the Google teams to look at different experiences of others using Kubernetes and how to optimise the parameters for autoscaling,” says Reda. “We receive excellent support from the Google Paris team. If we identify any issues, they always have a way to fix it.”
Scaling for future growth
Using Google Cloud Platform, Streamroot is now easily able to process over 40,000 https requests it receives every second. Thanks to autoscaling, fewer human resources are needed to manage the platform overnight, and the team has more confidence in its ability to meet peaks in demand. “With autoscaling, the work is mostly tuning and configuration, so at night there is less pressure on our backend teams,” says Reda. “We sleep better.”
Now, Streamroot is considering using Google Cloud Platform as part of its data pipeline, using Google Cloud Dataflow to analyse and improve its peer-to-peer algorithms. It is also planning to extend its multizone usage to get even closer to its customers around the world, and further improve its service.
“In one word, we’re facing the best kind of challenge: growth. That’s the one challenge that any start-up would want to have, and we’re very happy to partner with Google to meet it.”—Erica Beavers, Head of Marketing and Partnerships, Streamroot
The next challenge is the 2018 World Cup. “We’ve prepared our backend capacity, agility, and ability to scale our services to the traffic load that our customers are going to have,” says Erica. “This is one of the biggest events of 2018, for us and for our customers, and we trust Google to support us in delivering top-notch service worldwide.”
“Beyond that, we are looking towards the Olympics and continuing to scale our platform to hundreds of millions of users at the same time,” Erica continues. “In one word, we're facing the best kind of challenge: growth. That’s the one challenge that any start-up would want to have, and we’re very happy to partner with Google to meet it.”
After more than six months in production, Streamroot finds Kubernetes on Google Cloud Platform an amazing tool. The company has shared some of its experiences and lessons learned on the Streamroot Blog. If you are interested in meeting Streamroot, its team members regularly attend the Cloud Native Computing Foundation meetups in Paris.
Streamroot uses peer-to-peer networks to enable high-quality online video delivery for live streaming and video on-demand (VoD).