Infrastructure requirements for real-time applications

›

Defining the hosting needs of real-time applications

Deniz Arslan

Sales Manager

23 February 2024

Real-time applications (RTA) have specific hosting needs such as reliability and high performance. In this article, we look at the main pillars of real-time applications and analyze their hardware needs to establish why it is important for service providers to focus on low latency, reliability of service and high performance to ensure their end users get the experience they have paid for.

Not all hosting use cases are equal. This is painfully obvious to most, but especially those who work with services that rely on reliable and high-performance connections. Performance is a key non-negotiable in real-time applications as well, and this is why it is important to contextualize what this entails for the many use cases and types of services within the real-time ambit. But before we take a deep dive into issues of quality and low latency, it is first important to define real-time applications and the various types of services.

Defining RTA

Think of a website you use often, such as a news site or something to store data online. The data is stored on a server and is transmitted to you when needed.

Conversely in RTA, the data is transmitted between two points and needs to be received in real time to ensure that users of the service can benefit from immediate transmission. Use cases for this include voice and video calls online on your favorite social media services or streaming a live sports game online. There are a host of other use cases, but we will delve deeper into these a little later.

The three pillars of RTA

Interaction

Interaction, one of the three RTA pillars relates to audiences or users interacting with each other in real-time. This includes all voice and video conferencing tools or chatting applications where information and data packets get sent from one endpoint to the other. This information can naturally not be stored in a database somewhere as the data is being created on the fly and needs to be transmitted to the other side as fast as possible so that the conversation or interaction is organic and understandable.

Distribution

Distribution-related RT applications are related to the dispensation of on-demand video live video. This includesI also streaming and broadcasting use cases. Both live and recorded content fit into the ambit of distribution RTA. Examples of this include both streaming applications YouTube and Netflix, and on the live side, streaming any sports match online — football, tennis, or any other sport for that matter. With the rise of smart televisions, the use of cable TV is steadily being overtaken by the use of streaming applications on the TV as well, which means that the user base for online streaming in sports is steadily increasing.

Production

The production pillar in real-time is very different from the first two, in both infrastructure requirements and the scope of work. Production real-time environments is related to both game and video production. The gaming, filming and television/streaming show industries rely heavily on real-time production environments to manage teams and tasks globally. From making working possible through remote workstations and cross-border collaborations for large teams to CGI production, faster rendering, AI development and more, both GPU and CPU servers play a core role in producing the quality content we consume and interact with online. This includes games, films, shows and a lot more.

Key infrastructure considerations for interaction-based RTA

Interaction RT applications naturally require immediate transmission, and thus heavily depend on the bandwidth connection that the user has or the accessibility of the Internet to that user at that given moment. The quality of the connection is generally determined by the Internet Service Provider of each user, but this alone is not the deciding factor on the quality of service. The ISP’s connection to the rest of the global network and the quality and number of their peering and transit relationships is an added factor to consider.

Since this is normally a two-way connection street, the connection on both sides must be good enough for both user groups to be able to communicate with each other with minimal latency to avoid stutters, drops in quality, or disconnections.

The distribution pillar and factors to consider

The distribution pillar can get more complex based on the service you are running, and whether the transmission is live or recorded. For recorded transmissions, the distributor or content host can use a database to store content, but their end users will still have to be able to stream that content onto their own devices.

Live broadcasts and added complications

The complications add up when the content being streamed is a live broadcast. Sports or news broadcasters are already familiar with the myriad challenges that tend to present themselves when posting live content online. The content first has to be shot in real-time on a camera and transmitted back to a production van or studio to add an overlay of graphics (logos, score updated, tickers at the bottom and more) after which it needs to be sent to the nearest point of presence for disbursal to the audience. Depending on how fragmented the audience is and how popular the content is, this broadcast might even have to be sent out globally.

It's about catching those unmissable moments

Any sport or live transmission is live and watched as it happens because it moment is considered priceless by the audience. Whether that a match where a goal could be scored at any moment or some breaking news that has to be kept up with minute by minute, each moment is significant and its value lies in the present.

For the broadcasters themselves, having an audience that gets what it wants is crucial to retain viewership. If the platform is on a subscription-based model, quality will be one of the key factors of consideration for anyone who is committing to a medium- to long-term monthly payment model. And if the service is based on advertisement revenue, quality helps with advertisers as well.

The infrastructure layer matters

While building a streaming application is easy, the quality of content on it is heavily dependent on the infrastructure that supports it. This means that the servers that process the video to connections from where the application is hosted, and those to the final consumer, are all important.

Key infrastructure considerations for production-based RTA

Compared to distribution and interaction, production real-time applications are a whole different ball game. If we look at the production process for a film as an example of what might be needed, a film with a standard runtime of over 2+ hours requires a lot of work and hours put in to get the final product out.

In pre-production you need conceptualization exercises, there are rounds of back and forth over creating the storyboard and a lot of artwork and visuals are being sent back and forth before the team can be ready for filming. In today’s context, this entails cross-functional teams sending gigabytes of files across the global network, and high-speed transfers are needed to ensure that the workflow is not interrupted.

Real power is needed in the production phase

Producing a film requires more than just filming. Computer-generated imagery is a core part of today’s movie industry, and computing does a lot of the heavy lifting. Of course, there is also the rendering of 4K files for the many versions that will be considered, until the editing is finalized. There are films for instance, that could require 30 million computational hours to render. Remote workstations are also a core asset in this process. GPU servers located away from the place of work can handle the bulk of the workload without having to deploy multiple heavy and expensive machines in various office locations.

High-performance requirements across the stack

Apart from servers and computers that can work on producing a quality final product, a high-performance network that can move large-sized files is also necessary. Another consideration would be to perhaps centralize the files so that they are easily available for everyone when needed.

Final verdict: Handle the micro and macro with one provider

i3D.net specializes in high-performance workloads, with over two decades of experience in handling game development studios, publishers and enterprise clients. Much like film, broadcasting and other real-time applications of compute and graphic workloads, video games also rely on speed and performance for users to enjoy the experience. What it boils down to at the end is the network that connects all aspects, from production to distribution and consumption and the infrastructure that runs everything. Low latency is a core necessity, as is a partner that truly understands the value of delivering the product to the users as fast and as reliably as possible.

Main Take-Aways

This article looks at the main pillars of real-time applications and assesses their hardware and infrastructure needs.

Compute

Cloud

Related Pages

Game Online Services

Related Pages

Connectivity

Related Pages

CDN

Related Pages

Colocation

Related Pages

By Application

Compute

Low Latency Connectivity

Content Delivery

DDoS Mitigation

Defining the hosting needs of real-time applications

Table of Contents

Defining RTA

The three pillars of RTA

Key infrastructure considerations for interaction-based RTA

The distribution pillar and factors to consider

Key infrastructure considerations for production-based RTA

Final verdict: Handle the micro and macro with one provider

Main Take-Aways

Related Links

Need help from our Customer Support Team?

Sales

Features

Solutions

Products

Company