Long reads


Q&A: Alexey Petrovskikh, G-Core Labs on low latency streaming

Alekey Petrovskikh, head of streaming platforms at G-Core Labs, talks to DTVE about meeting the challenge of ultra low-latency streaming.

As live streaming grows in popularity, delivering ultra-low latency is becoming critical. What is the G-Core Labs streaming platform and how can it help to solve the challenge mentioned above?

The G-Core Labs streaming platform is a unique turnkey solution for live video and video-on-demand streaming that can be used to deliver video content to any device anywhere in the world in Full HD and 4K quality. We understand that viewers want to watch the decisive moments of broadcasts without delay and get access to videos and films that interest them at the moment they press play. That’s why we ensured a video delivery time of one or four seconds. By default, we support a delay of up to four seconds for all video streams. Such a broadcast can be conducted to an audience of 100 million viewers. Customers can also subscribe to the ultra-low latency streaming service – a premium platform service where, with the help of the WebRTC technology, the video stream delivery time is minimised to one second.

Using our streaming platform, users can quickly establish a full-fledged online broadcast of a TV channel from scratch, or launch an online cinema that airs films, series, sports matches, poker tournaments, rock concerts, marketing videos, and personal video blogs. Everything can be set up from a personal account or API. Customers don’t need their own transcoding servers, CDNs or broadcasting engineers.

What other ways of organising broadcasting on the Internet are there? Please explain how your solution is different from YouTube and Facebook streaming services?

It’s very easy to stream nowadays: take out your phone and send a video stream to YouTube or Instagram. However, the website can block your account for violating any of its terms of service, or the broadcast may be interrupted, and the support service will reply to your email address the next day or a week later.

For stable and cost-effective broadcasts for ten, thousand, hundred thousand or a million viewers, it’s better to use proven solutions. Few people will be happy if, during the speech of the company’s CEO, the broadcast stops for ‘unforeseen circumstances’.

We optimise our platform for the needs of the customer. You don’t have to pay for extra terabytes of unclear traffic or other hidden modules and fees.

If necessary, the video stream can be relayed to several social networks at the same time by pressing a couple of buttons. This way you can increase the audience of your video 10 times .

Unlike YouTube, all video content can be protected against unauthorised viewing. In addition, the platform can be used as an all-in-one solution for broadcasting or to partially implement, for example, only transcoding, CDN delivery, or the HTML5 player.

How did you manage to achieve low-latency streaming? What steps did you take?

Video buffering and content unavailability annoys viewers, but they don’t realise how much it costs to support instant viewing of high-quality video. Setting up high-tech streaming by investing in your infrastructure is always very expensive.

The new standards for low-latency-streaming based on HTTP (DASH, HLS) – which our platform also uses – are based on the use of a variant of the HTTP 1.1 protocol called Chunked Transfer Encoding. Thanks to this, it is possible to transfer new parts of the last (live) video segment to the customer immediately as they arrive from the transcoding servers.

At the same time, it’s necessary to ensure the immediate transfer of Chunked Transfer Encoding through the content delivery network all the way from the source to the customer without creating unnecessary traffic at the source.

We’ve successfully integrated these technologies into our platform. As a result, you can use the full power of G-Core Labs content delivery network (100+ points of presents on five continents) to deliver traffic while still delivering very low latency video.

What other key features are important in a streaming platform to meet the evolving needs of publishers and content service providers?

An important functionality of our streaming platform is reliable protection of customer content against copying. To protect Live broadcasts, as well as DVR and VOD recordings, we use various anti-copying methods that restrict unauthorised user access and permit streaming only through the customer’s resource, such as AES 128/256 Encryption, Tokenisation, and Signed URLs. We support the highest level of DRM protection.

What software and hardware does your streaming platform use, and why?

Our hardware infrastructure for video solutions consists of several geo-distributed clusters for receiving, processing, and delivering video to the servers of G-Core Labs own content delivery network. Some of these servers are used for CPU transcoding, and others for GPU transcoding. The standard configuration of servers includes Intel Xeon Gold 6152 and 5220R processors, SSD and HDD drives with triple replication, a High IOPS SSD for applications requiring high IOPS, and a KVM hypervisor.

Low latency software is our own solution in combination with a number of interchangeable open-source solutions for receiving streaming video, its transcoding, and distribution.

What key future trends in video streaming do you anticipate?

The media and entertainment segments based on live streaming or on-demand delivery are actively growing. According to a number of analytical agencies,  in the next 2–3 years video will account for about 90 % of all Internet traffic. In the next 1–3 years, industries such as video games, esports, TV and media will fully switch to streaming with minimal latency . Online education, telemedicine, VR/AR, financial analytics, and e-commerce are already using premium real-time video streaming.

As the quality and speed of video delivery to any device grow, advertising budgets are being redistributed in favour of video marketing.

Solutions for audience retention and content monetisation based on artificial intelligence are actively appearing on the market. Today, G-Core Labs is developing AI-based products that allow an internet user or a TV viewer to receive information in one click about the actors and landmarks in the frame, or the dress or watch an anchor is wearing. This will provide great new opportunities for advertisers.

We also expect that big sporting events will be held in 8K in the near future, and that 5G mobile technologies will deliver this video content to a large number of mobile devices.

Here’s another trend: in a number of markets, including the USA and China, the production of 12K content is being developed today. This is in particularly high-demand in the VR and AR segments. This also raises the challenge of high-quality and seamless delivery of data-heavy traffic to end users and viewers around the world.

All this translates to the investment of more time and resources in video streaming.

This is sponsored content. 

Tags: G-Core Labs