ReSTful APIs make it easy for applications to manage data and metadata on a global basis without intense recoding of applications.
One of the most important concepts we haven’t yet addressed here in our Red Hat Perspectives series on software-defined storage is simplicity, and the first place it shows up is in the comparison among file storage, block storage, and object storage architectures.
File storage is a hierarchical structure in which each file is identified by describing the path to it -- what drive it is on, in what subdirectory, under what file name and extension, etc. Metadata is stored in the file system describing who can access each file and other specifics. As more users grow to petabytes of data and millions of files, file storage systems have proven not to scale very well. This can become intricate and complex.
Block storage leaves the identification to the application and stores unidentifiable blocks of data with no metadata. Everything is tracked, located, and combined into files and called by the application, leaving total responsibility in the hands of developers. Since latency increases with block storage as the distance between the application and the storage increases, most block storage use cases tend to remain localized.
The Simplicity Of Object Storage
Objects are data usually in the form of files that have been bundled with all of their metadata into a container (sometimes referred to as a bucket), including such items as associated application, importance, and required security level.
What keeps object storage simple is that one of the primary ways to access it uses ReSTful application programming interfaces. No, this does not mean that using them quiets the mind and calms the soul. Representational State Transfer (ReST) uses simple hypertext transport protocol (HTTP) commands such as PUT, GET, and DELETE. The bonus inherent in using HTTP is that it was built to travel the Internet, making long distances between applications and data inconsequential. ReSTful APIs make it easy for applications to manage data and metadata on a global basis without intense recoding of applications. Simple!
Using these industry standard APIs makes application migration and integration seamless. The tremendous simplicity of HTTP-based ReSTful APIs makes them far more scalable, too, which is why object storage is becoming so popular as more and more organizations grow to petabyte-scale data storage.
Simply, Reliably Accessible
In part because of its simplicity, object storage with ReSTful APIs is accessible from such industry-standard platforms as the appropriately named Amazon Simple Storage Service (S3), the Storage Networking Industry Association’s Cloud Data Management Interface, Google Cloud Platform, OpenStack Swift, and other native protocols. ReST is simply not playing hard to get.
One advantage of using software-defined platforms such as Ceph or OpenStack Swift is that they enable users to build solutions on commodity server hardware and disk drives and replicate objects to reduce latency and errors that can be introduced by long-haul transfers of data.
All of the platforms that support object storage, including Amazon S3, provide ReST APIs that make it easy for programmers to work with objects and containers.
Simply put, ReST basically amounts to using simple, familiar Web protocols to facilitate moving data into and out of applications. To learn more about ReST and how it is facilitated by products such as Ceph Storage, visit Red Hat at www.redhat.com/en/technologies/storage/ceph .
Steve has more than 15 years' experience in product marketing and management serving enterprise customers working for industry leaders such as Red Hat, NetApp, Violin Memory, and HP. He has travelled around the world in his career and launched numerous storage and ...View Full Bio