What’s Under the Hood: Technologies We Use

Max Lapshin
2 min readJan 14, 2020

--

We are making solutions for storage, delivery, and optimization of the video stream, and these are quite complex tasks that require the most advanced technologies. I would like to tell how exactly everything is arranged with our company.

Let’s start with programming, this is the basis of our work. Of all the programming languages, we use Erlang (for all network, protocol, format and container handling), C (for transcoding) and Rust (for IP camera firmware). Rust turned out to be the best embedded solution for us because it gives just enormous safety with fantastic performance. Just an example: our on-camera streamer hasn’t ever segfaulted.

Using standard protocols and interfaces, like WebRTC, we arranged direct publishing of videos from webcams, and we managed to achieve maximum bandwidth from the channel. We also wrote an adaptive bitrate adjustment algorithm that checks and tries to raise the bitrate until errors begin to occur. Moreover, we’ve set up the WebRTC in such a way that the cameras can be accessed directly, in peer-to-peer mode. At the moment, in the world, there are very few cameras that can do this, so here we are at the forefront of technological progress.

We use neural networks as well, and we develop them ourselves on the basis of some ready-made basic tools. The neural networks that we develop demonstrate the maximum performance with the minimum resources spent. Our team has mathematicians with a scientific degree that are responsible for developing models for neural networks. The development of mathematical models is a rather specific area where many things need to be taken into account and where complex mathematical calculations are inevitable. Our specialists solve complex problems that require a very deep understanding of the topic.

We also recently started to develop our own hardware, and this has given us a serious boost in power consumption. We’ve reached the power acceleration of 20 kW per hour for 50 channels, and this is about 5–6 times better performance compared to our competitors. We started selling our hardware transcoder last fall, and this was a great breakthrough for us.

We are always looking for the best technologies and solutions that can help our work. In addition, practice shows that we can also take basic, standard technologies, and find ways to use all their capabilities to make our customers experience better.

--

--

Max Lapshin
Max Lapshin

Written by Max Lapshin

Board Member of Otter Video — all-in-one solution for OTT delivery. Founder and CTO at Flussonic — video streaming solutions.

No responses yet