Socceroos' Mile Jedinak scores against France during the 2018 FIFA World Cup
Key technology partnerships played a crucial role in helping TV broadcaster SBS provide the full 2018 FIFA World Cup user experience across multiple devices.
It was no mean feat, particularly surrounding the streaming woes that kicked off as Australians tried in angst to watch some of the patchy coverage initially through Optus Sport.
Eventually SBS took over handling the broadcast of the entire tournament after Optus admitted that it couldn’t handle the capacity that such an event demanded from Australian viewers.
In the end, more than 9.1 million Australians watched the World Cup coverage, with SBS clocking up about 7.8 million live streams across video and radio, out of which 3.6 million was video along with 4.2 million audio streams, supporting 12 different languages.
At its peak, there were about 280,000 simultaneous users streaming live video during the Australia vs. Denmark match.
“That’s a substantial load on all the systems,” said Jeremy Kelaher, solution architect for enterprise digital services at SBS, when speaking during the New Relic FutureStack event in Sydney.
"We spent a lot of time measuring and our team was continuously monitoring everything to flag any potential issues that could impact the audience. Our teams and partners knew what to look for, and they were on-standby to adapt to any situation."
One poignant factor Kelaher pointed out was the importance SBS placed in its technology partnerships to help provide the most interactive viewer experience of the tournament across multiple devices, considering the size of its 40-strong IT team.
The broadcaster worked with a range of partners to help deliver greater stability and scalability as efficiently as possible.
In its preparation leading up to the tournament, SBS upgraded some of its technologies to help scale its infrastructure, relying on key partnerships with Amazon Web Services (AWS) using cloud services such as EC2; and upgrading to Aurora DB and Elastic Search along with using AWS Elemental - transcoding and encoding.
Deluxe was used for IP playout; the media asset management system was based on Dalet; along with Akamai for the content delivery network, which also provides DDoS protection; New Relic for monitoring; and Janrain for customer identity and access management.
NGNIX also played a core part of SBS’ architecture, with the broadcaster also leveraging Broadcast Australia, Telstra and Optus Satellite to help distribute content to its multilingual audience.
Kelaher explained that getting the most of out its army of tools such as New Relic, was also an important factor, particularly when it came to video quality of service that provided a detailed diagnosis on the sorts of problems that could interrupt a user’s viewing experience.
"During an event like the World Cup, there’s continual conversations with our partners to ensure we can scale up as needed and although we have to manage our relationship with these partners, the great thing is that they take a lot of the load off SBS,” he said.
"We don’t have an enormous tech team, and consequently there’s no doubt that we need partners to help us. One of the best things about cloud is that you can get those best of breed partners and they can enhance your team."
Kelaher said in the lead up to the tournament, SBS conducted numerous testing in particular with the pre-scaling of its video APIs and content repository, adding memcached layers as well as considerable load testing.
This was in addition to monitoring to ensure it could handle the extreme spikes that were expected, alongside producing a heat map of key scaling points, tests and current capacity and scaling time.
"When you’re experiencing a busy event, you need to be aware of what is going to happen the following day, and need to be aware of scaling time, and if that is not scaling as fast as you’d expect, then you know you have a problem,” he said.
The broadcaster kept tight tabs on when all the matches were occurring, when it thought the busiest times were going to be, and where it thought difficulties may potentially arise.
“We needed to keep everyone in the loop, so they could have their support staff ready to go,” he said.
Some of the key lessons SBS took away from the tournament broadcast included having good monitoring and data management policies, as well as not assuming knowledge with the audience. In other words, be prepared for surprises, Kelaher said.
“We gave our users a lot of options on how to consume our content, but one of the most popular experiences was through the iPad and Web iOS is a hard environment for complex web experiences,” he said.
“Users with old iPads, lots of apps running, or devices with little memory, created some more support issues that any other platform, but fortunately we had our contingency plans in place.
"Good performance data management policies are essential. We generated a ridiculous amount of data, and if you don’t have good data performance management policies - then you’re not going to be able to look back in retrospect and see exactly what happened."
