Cloud Computing gave us several changes in how we handle IT nowadays. Common tasks that used to take a lot of time received great automation and much more is still about to come. Another interesting development is the “Software defined X”. This basically means that infrastructure elements receive larger automation as well, which ends up being more scale able and better to utilize from applications. A frequent term used lately is the “Software defined Networking” approach, however, there is another one that sounds promising, especially for Cloud Computing and Big Data: Software defined Storage.
Software defined Storage gives us the promise to abstract the way how we use storage. This is especially useful for large scale systems, as no one really wants to care about how to distribute the content to different servers. This should basically be opaque to end-users (software developers). For instance, if you are using a storage system for your website, you want to have an API like Amazon’s S3. there is no need to worry about on which physical machine your files are stored – you just specify the desired region. The back-end system (in this case, Amazon S3) takes care of that.

As of the architecture, you simply communicate with the abstraction layer, that takes care of the distribution, redundancy and other factors.
At present, there are several systems available that takes care of that: next to the well-know systems such as Amazon S3, there are also other solutions such as the Hadoop Distributed File System (HDFS) or GlusterFS.
 
Header Image Copyright: nyuhuhuu. Licensed under the Creative Commons 2.0.