A joint venture formed by Groupe M6 and RTL Group (Bertelsman), Bedrock Streaming provides streaming platforms for video on demand (VOD) services with or without subscription. Historically, its platform started within the M6 Group, serving a single client. Today, the company operates the services of several clients, including 6Play and Salto in France, Videoland in the Netherlands, RTL Play in Croatia and Belgium and RTL Most in Hungary. This development of activities has led the company to evolve its infrastructures to go into the cloud. In this context, monitoring has become more complex, with an increased need for visibility on both the back-end and the front-end, which includes the Web, mobile devices and smart TVs. To meet this need, Bedrock chose to implement New Relic’s observability solution.
“Initially, our platform was hosted in two datacenters, with a single-instance, single-tenant system. To accelerate our development and serve several customers, we moved to the AWS cloud in 2019,” says Olivier Mansour, deputy technical director at Bedrock Streaming. The company then switched to a multi-instance and multi-tenant architecture, to adapt to the needs of its different customers. Application monitoring was then performed with a set of open source tools, including Elastic’s ELK suite, Prometheus, Grafana, Graphite and StatsD software. These solutions worked very well in the initial BtoC context, where the teams had good visibility on the evolution of the platform and a precise schedule, making it possible to anticipate investments on a yearly scale. But the conditions were no longer the same in a BtoBtoC model, with increasing amounts of data. “Today we have three major platforms in production and 75 microservices. On these environments, we need to have scalable monitoring and relevant indicators, with a vision both by instance and by customer,” explains the CTO. In addition, the company has a strong DevOps culture, with 300 employees working on the development and technical operation of its streaming platform. The monitoring solution had to be suitable for all teams and adapt to the diversity of services and technologies in place, in order to be able to monitor and analyze events across the entire chain, and facilitate remediation in the event of dysfunction.