How to choose a video conferencing server

Choosing a server is a task that everyone who is involved in building or scaling a video conferencing network (or, in short, video conferencing) faces in one way or another. Let's try to make out the main issues that arise when choosing the infrastructural elements of a video conferencing system.

Architecture of the VKS system

As many know, the classic video conferencing system is built on the basis of a multipoint video communication server (MCU). Such a server accepts video streams coming from end devices (terminals), and performs all the necessary operations for the formation of images that will be sent to other terminals. This transcoding is the conversion of video from one format to another, transrating - changing the bit rate at which the image is transmitted, as well as mixing. The last operation involves the layout of images in accordance with the layout that is convenient for a specific user.

The last few years on the market are actively promoting alternative solutions with simplified MCU servers. Such servers simply forward video streams between end terminals — this feature is often called proxying. They are not able to carry out transcoding or transrating, some servers can perform only basic mixing functions. In this case, the main load on the processing and the formation of the “picture” falls on the final terminals.

One of the technologies that allows the use of a simplified server is layer-by-layer video encoding on the terminal. Each layer increases the resolution of the image, so when transferring video to another terminal, the server does not need to recode anything - just select so many layers so that the resolution matches the characteristics of this terminal and the communication channel with it.

The terminal can also prepare full-fledged video streams with different resolutions (for example, CIF, SD, HD ...). In this case, the server task also comes down to choosing the desired stream to send to another terminal. With such schemes, the MCU server becomes, in effect, a video stream router.

So, there are two options: a full-fledged MCU with centralized video processing and a simplified server (proxying) with transferring most of the workload on video processing to terminals. Which option is better? In most cases, a centralized server is definitely preferable. This solution greatly simplifies management and scaling, reduces the load on communication channels and end terminals, provides flexibility in the implementation of various layouts, including individual ones for a particular user. In addition, the availability of a centralized full-fledged MCU server simplifies the integration of the video conferencing system with other corporate systems. Fault tolerance of a centralized solution can always be improved by installing multiple servers and combining them into a cluster, which can also be geographically distributed.

The option with a proxy server and distributed video processing may be acceptable for certain situations. However, it has more disadvantages than advantages. These are the need for powerful terminals capable of video conversion, higher load on the channels, less flexibility, possible difficulties with controllability and integration with other systems. Very often, solutions with a proxy server were developed for webinars and video chats (I may tell you about the differences between the latter and full-fledged video conferencing), attempts to adapt them to video conferencing are not always successful. When using such solutions, users may encounter problems associated with high channel load and / or low video quality.

MCU implementation

Critics of using full-fledged MCU as their main disadvantage is called high cost. This stereotype that took shape decades ago has long been outdated and has ceased to be relevant. But, unfortunately, even experts still sometimes confuse the architecture of video conferencing systems with the specific implementation of the MCU server. Five or ten years ago, to perform demanding video processing operations (transcoding, transrating, mixing) really expensive hardware was required. In addition, there were few offers on the market, weak competition, which did not contribute to lower prices.

However, over the past years, technical progress has led to a significant increase in computational and graphical capacities of even inexpensive servers of standard architecture. Therefore, the full functionality of the MCU today is effectively implemented on the basis of standard server platforms with available graphics processors. Neither expensive DSP processors nor other complex hardware elements are required. All this allowed manufacturers to offer really effective full-fledged MCUs at a relatively low price. Moreover, additional infrastructure elements, such as a registration server, a recording server, and video broadcasts can be implemented as additional software modules that are installed on the same server. Separate physical servers are no longer required for registration, recording and broadcasting, which also reduces the cost of a complete solution for the video conferencing system.

Even if you are offered today an expensive MCU server, made in the form of a specialized device, it may turn out that it has an ordinary x86 computer under the hood. And such a computer is worth an order of magnitude less than the amount requested for the so-called “specialized device”.

Repeat an important point. Regardless of the system architecture (centralized or distributed), the server implementation may be different: specialized hardware and software platform or software for a standard server, including for work in a virtualized environment. Do not believe that an expensive specialized hardware and software complex is necessary for a centralized system as a full-fledged MCU, and for a distributed system, fairly inexpensive software. It is a myth. There are effective full-fledged MCUs on the market, made in the form of software for a standard server platform.

Compatibility

In recent years, significant changes have occurred in the VKS market. A large number of mergers and purchases has led to the fact that many well-known brands have disappeared from the market, product lines have undergone a serious audit. Here are just a few examples. In 2008, the well-known Italian manufacturer of video conferencing systems Aethra was acquired by Israeli Radvision, which subsequently, in 2012, bought by Avaya. But this is not the end. Having retained the Scopia product line, Avaya in 2014 resold the main part of Radvision to Spirent Communications. The ClearOne company, which in 2012 acquired the old-timer of the VCON video conferencing market, in 2014 also bought the developer of cloud-based video conferencing solutions, Spontania. Once upon a time, one of the top three leaders in the VCS market, Lifesize virtually ceased terminal development, separated from Logitech, and focused entirely on cloud infrastructure and software. Finally, Polycom, almost immediately after a failed deal with Mitel, was bought by Siris Capital.

What does this mean for the customer? And the fact that in the conditions of such turbulence in the market when choosing a key infrastructure element, which is the MCU server, guaranteed compatibility with a large number of products from other manufacturers is necessary. Note that the leaders of the VCS market, when developing their servers, are naturally oriented, first of all, towards the support of their terminals. Therefore, smaller companies are usually able to offer broader compatibility, which are initially focused on customers with a “zoo” of equipment when developing their products. Their solutions can be the most effective integrating element, which will ensure both the operation of the existing equipment and the possibility of further scaling.

Even if you use mono-vendor solutions of a very large manufacturer and are absolutely sure that it will exist forever, anyway, support by the north MCU of solutions from other manufacturers may be in demand. After all, the market turbulence applies not only to manufacturers of video conferencing systems, but also to companies from other areas. This means that your company may also be involved in the M & A scheme (from the English mergers and acquisitions). If tomorrow your top managers decide to purchase another company, and it will have a VCS system from the wrong manufacturer you are working with, you will face a difficult integration task. Its solution may require additional costs and lengthy testing, and it’s not a fact that the corresponding funds will be provided for in the budget.

If we talk about compatibility, then of course, first of all, we need to pay attention to the support of the main protocols and codecs, including the H.323, SIP, WebRTC signaling protocols, as well as the H.261, H.263, H.264, H codecs. 264 High Profile (AVC & SVC), VP8, H.265 (HEVC), etc. But often this is not enough. It is advisable to request from the supplier the test results of its server with other servers and terminals. It would also be helpful to request information on the use of this supplier’s products in multi-vendor projects in other companies.

Webrtc

Speaking about changes in the market, it should be noted that VCS has already gone beyond the meeting rooms: an ever wider circle of corporate users is using video to interact with colleagues, partners and customers from their workplaces and even from mobile devices. The number of potential video conferencing terminals has dramatically increased due to the development of WebRTC technology, which allows participating in video conferencing sessions from a browser without having to install any additional software or download a plug-in. Today in the world there are more than a billion computers with browsers that support the WebRTC technology.

Accordingly, an important requirement for the MCU server is to support this technology. In this case, the MCU server can be an effective gateway for interfacing a classic video conferencing system that combines group terminals that are installed in meeting rooms with a large number of personal video conferencing terminals, which use computers, tablets and smartphones with WebRTC support.

Concluding the discussion on compatibility and integration, it is important to note that the matter is not limited to supporting various VCS environments. In many projects, there is a need to integrate a video conferencing system with a corporate telephony system, usually via IP-PBX. Ensuring such integration is an important feature of the VCS server.

Security

VKS is often used to conduct important business negotiations, so ensuring a high level of security is extremely important. By the way, because of concerns related to security, customers are in no hurry to use cloud services, preferring to have a video conferencing system fully at their disposal. It is desirable that such a server has a built-in firewall, and the procedures for connecting to the conference provided for reliable authentication algorithms.

Recently, security issues are increasingly associated with the origin of the product. It is clear that “other things being equal”, it is better to give preference to domestic development - in addition to “sanction-resistant technologies”, you will receive high-quality local support from specialists who are well acquainted with the realities of Russian projects.

Tight security measures can, in turn, cause problems when organizing video calls. This, above all, is about the problems in passing calls through firewalls and address translation systems (NAT) installed on the perimeter of many corporate networks. Technologies to circumvent such obstacles are already run in (NAT H.460.1, Firewall / NAT traversal), it is enough to make sure that your chosen server supports them.

Quality

The quality of the final picture that the user sees is determined by a lot of factors, including the quality of the program code and the hardware elements of the system. Today, it’s not surprising anyone to support VCS systems with a wide range of resolutions: from QCIF to Full HD and even 4K.
But in order to ensure such a picture quality, not only the technical capabilities of the terminals and the MCU server are important, but also the capacity and quality of the communication channels. With this in Russia, alas, not all is well. Therefore, it is extremely important that the VCS server supports efficient QoS mechanisms, including traffic prioritization, as well as the ability to work on packet-loss channels.

It is desirable that the server supports technology to compensate for packet loss: up to 1% of packets without noticeable degradation and up to 5% of packets without critical degradation of image quality. With a further deterioration in quality and / or a decrease in the capacity of the communication channel, the most critical flows should be supported, first of all, those that carry audio information. Even when a video is lost, the voice can always convey the necessary information, so voice support is the most critical.

findings

The most progressive and reliable VKS servers today are software solutions implemented on the basis of standard architecture servers. They give the optimum ratio of manufacturability-quality-reliability-cost. Such platforms are as flexible as possible - they are easily customized and adapt to the needs of any organization. When choosing among software platforms, we pay attention to the technical characteristics of the hardware, the compatibility of the platform with solutions that are already installed in you or in organizations with which you will communicate via video. Support for WebRTC is also not superfluous.

I will add that Russian-made software servers, other things being equal, are cheaper, and, in the current political and economic conditions, give more confidence in the future. I hope these recommendations will be useful and make life easier for many! I am pleased to answer all the questions arising from reading.