Your browser is not supported. Please update it.

10 September 2020

Cloud-native, a short-term buzz or the next evolution of video?

Faultline’s buzzterm detection engine has been working in overdrive throughout 2020 as technical and marketing terms beguile the video industry into a false sense of security. Among all the “super aggregators” and the “new normals”, a late case is being made for “cloud-native” to claim this year’s buzzterm crown. With cloud-native discussions featuring heavily during the virtual IBC Showcase, we set out on a mission to clear up what the latest cloud-native hype is all about.

Of course, being cloud-native is not a direct product of the pandemic by any stretch, with Faultline first registering the term’s use in 2016 in an article covering Cisco’s acquisition of cloud security firm CloudLock. This came some two years behind our sister service Wireless Watch, where cloud-native mentions have ramped up massively in recent years, which is reflective of how emerging network technologies take time to be realized in the video community.

Certainly, global lockdowns have accelerated the deployment of applications, services and functions in public, private and hybrid clouds en masse, which has driven demand for technologies like microservices, containers, and serverless functions, using DevOps and orchestrated in Kubernetes, that make up your typical cloud-native application.

First, we must trace the opportunity back to the source of demand – operators. Tier 1 giants well along the path towards fully virtualized networks have identified a lack of cloud-native SaaS models for software licensing among the vendor community as a key technology gap. This is true for applications like cloud-native media creation platforms, as well as cloud-native wireless and cable networks themselves, where SaaS models can overcome the complexities of video using the latest technologies available, from the massive storage and processing savings that can be made by doing file-based transcoding in the cloud, to high-demand workloads like 360-degree video, where processing in the public cloud is miles more efficient. Disaster recovery is another main benefit of being cloud-native, slashing costs by removing the need to duplicate information across data centers, which gets particularly costly with double capex and double opex.

Honorable mentions of video technology companies being early to cloud-native platforms go to Harmonic, CableLabs, AWS, Ooyala, and Rakuten, while Sky Italia springs to mind as a solid operator example. Sky Italia’s linear video streaming platform is based on Kubernetes and microservices to serve event-based and 24/7 channels on OTT platforms, taking advantage of Harmonic’s cloud-native media processing software, including its content-aware encoding and VOS Cluster software which dynamically controls and optimizes on-prem and cloud video-processing resources.

Rakuten was first to deploy a fully virtualized, cloud-native network, initially for 4G and later 5G. Being something of an expert here, as a multifaceted company (not just a telco) using modern software-driven techniques to deploy new network technologies faster and cheaper than rivals, it was serendipitous to see Rakuten as half of a virtual IBC presentation arguing the case for cloud-native technologies, in this case cloud-native CDNs.

It was only last year when we heard one major CDN provider smear cloud-native CDNs as simply marketing campaigns built on removing costs, in terms of the lack of physical infrastructure that comprise traditional server-based CDNs.

A lot can change in a year and it has long been true that OTT players see the CDN as a commodity, while multi-CDN deployments are increasingly viewed as best practice for large scale content delivery. For companies wanting to design their own CDN, scaling is the hardest part in design, according to Devesh Gautan, Director of Rakuten’s Edge Platforms, MEC, Media Services and CDN segment.

Cloud-native CDN infrastructure can scale elastically to handle dynamic high-volume demand, and can be optimized for high throughput, low latency and smooth media delivery. This shortens time to market and optimizes hardware, by sharing hardware across different functions and scaling with demand, as well as reducing energy consumption and managing costs with virtualized infrastructure and decreasing capex on idle hardware. Rakuten has worked closely with Intel here.

Rakuten’s cloud-native CDN is aligned with upcoming 5G network roll outs – benefiting from autoscaling, auto healing, and customized hardware capable of delivering more data per server to support use cases like enhanced mobile broadband. Gautan added that a cloud-native approach can handle very high levels of concurrency, which will support massive machine type communications.

Cloud-native CDN technology can reduce or even eliminate the need for capacity planning and gives operators options to add their own CDN to existing network infrastructure. Something like the NEA-CDN 5 from Anevia, for example, which debuted at IBC 2019 for reducing network load by caching user requests and content to protect origin servers from multiple requests.

In short, harnessing cloud and virtualization technologies is the next evolution of CDNs to enable the decoupling of software from hardware.