Traditional storage relies on a file system interface – that is you present your network attached storage (NAS) as a file system interface such as a drive letter on Windows or as a volume on Mac or Linux. But traditional file systems have scalability issues – an upper limit on file numbers (typically millions) and as you approach this limit then performance becomes an issue. Object Storage does not rely on a file system to manage the content under its control and so, in theory, there is no upper limit on file numbers. This in turns allows you store petabytes and beyond of storage with no loss in performance. An object is in fact just a digital asset such as a media file.
And there are additional benefits. A file system does not allow you to store metadata with content – it’s a limitation. However, object storage does allow this making it searchable and intelligent about its own content without the need for a separate media asset management system. Or when using MAM’s they can be configured to archive and protect metadata as well as the media files themselves. It is also possible to manage different objects with different data management rules such as a number of copies and replication rules.
Facebook and Amazon S3 are built on Object Storage principles – Amazon S3 has trillions of objects under its management which it provides in an online cloud type environment. Object Storage must allow for loosely coupled inter-dependencies – in other words having different users and applications being able to access the content without impacting on each other. The MatrixStore architecture virtualises multiple nodes of physical hardware to enable it to be used securely by multiple users, and with minimal impact upon each other. Also, the flexibility of an object storage architecture means that multiple storage policies can be enforced, e.g., for how long objects are stored, how many instances of data are kept and the location of those objects.