The emergence of object storage as a mechanism to store digital video has modernized the way video content is stored and distributed. Whereas legacy systems that utilize a traditional POSIX file system architecture store video assets as files within a directory tree, object storage allows for an asset to be stored as a “data blob” in a potentially large namespace. This method of storing data introduces a number of advantages for digital content storage and distribution.
File systems have been around for decades. While many advancements have been made allowing them to scale, their general purpose architecture means that they’re designed to work for many different kinds of content, including databases made up of rapidly updated large files, logs that need constant updating and trimming, and various applications that need many of the legacy operations that are part of the POSIX specification.
Object storage, on the other hand, is a relatively modern method of storing and accessing data. While there are many different object storage platforms and protocols, typically all of them utilize a simplified command set that includes creation, reading, and deletion of objects. An object can be of any length, and typically is stored in a “flat” namespace with all other objects, only referenced by its unique name. In some cases, a single-level grouping of objects into namespaces is employed, typically called a “bucket” or “container”.
This paper discusses the requirements of modern IP-linear and cloud DVR systems that utilize object storage, the advantages and challenges with object storage, and how those challenges are met with hybrid object storage technology.