Concept - Video CDN [EN]

What is Video-on-Demand

Video-on-Demand (VoD) is a service that allows users to watch video content at any time they choose, instead of following a fixed broadcast schedule like traditional television. With VoD, viewers can select specific movies, TV shows, or other video materials, and the system delivers the chosen content instantly over the internet.

VoD technology relies on streaming data, which enables continuous playback without requiring the entire file to be downloaded first. This approach allows adaptive video quality based on the user’s internet connection speed, enhancing the viewing experience even on unstable networks.

VoD services are widely used in both the entertainment industry and corporate environments for training and presentations. Their popularity is rapidly growing due to convenient access, personalized recommendations, and the ability to watch content on multiple devices, from smartphones to smart TVs.

Large files delivery challenge

When it comes to delivering large files such as high-resolution video content, traditional Content Delivery Networks face several architectural and performance-related challenges.

First, caching large files (especially multi-gigabyte 4K videos) is inherently difficult. Storing such files efficiently often requires specialized logic at the web server level, since standard caching mechanisms struggle to handle partial caching or range requests effectively.

Additionally, large-scale caching demands significant disk resources. Using SSDs across the infrastructure becomes prohibitively expensive, while HDDs cannot provide the necessary throughput.

Moreover, naively caching every newly accessed file leads to inefficient cache churn. In large video libraries spanning hundreds or even thousands of terabytes, many files are requested infrequently. Storing them in cache immediately often results in premature eviction, long before the next request arrives. This wastes valuable storage and reduces overall cache hit rates.

Traditional Video Delivery Network architectures also suffer from limited traffic distribution flexibility. Content is typically served from the same node that received the request, which means peak demand must be met locally. As a result, providers must overprovision capacity at every node to account for potential spikes — an approach that scales cost linearly with traffic volume.