Scanrail - Fotolia
Object storage is a highly scalable system for organizing and storing data objects (files) without the use of a traditional file system structure. Object storage systems ingest data as objects with unique keys or IDs into a flat directory structure using simple commands. Metadata is stored with the objects rather than in an extensive hierarchical journal or tree. Search and retrieval is performed via key or ID searching.
In an object system , metadata is mostly stored with the object file and it can contain an almost unlimited amount of data about the data, which makes object storage systems much more efficient in terms of the way they're managed and archived.
It could be argued that object storage is a form of software-defined storage, but the challenge is that a definition of the latter has been hard to pin down and made even murkier by the marketing hype surrounding SDS. Software-defined storage, in theory, is a model for storage infrastructure in which storage services, traditionally provided in the form of value-add software placed on an array controller, are moved off of array controllers and into a software layer that operates as part of the server hypervisor software stack.
While software-defined storage is touted as new and revolutionary, it is actually very much akin to how storage services were delivered in IBM mainframes using System Managed Storage (SMS) beginning in the late 1970s. At that time, the function of SMS was very similar to the current description of the functionality of object storage systems in that SMS enabled the management of data onto primary storage, and its later migration to less expensive storage (hierarchical storage management functionality), across virtually any commodity storage operating under the domain of the object system.
The key difference between object storage systems and software-defined storage today is that the latter typically provides storage only for data emanating from workload that has been virtualized using a specific brand of server hypervisor software. VMware and other hypervisor vendors appear to be seeking to create isolated storage repositories dedicated only to their own virtual machines and exclude the placement of other hypervisor workload data (or non-virtualized application data) in the dedicated storage. In my opinion, this is a strategy in sharp contrast to most object storage products, which are geared toward creating a unified, universal platform.
Object storage systems making smart storage smarter
How does object storage perform in the cloud?
Dig Deeper on Storage architecture and strategy
Related Q&A from Jon Toigo
Cache memory and RAM both place data closer to the processor to reduce latency in response times. Learn why one can be faster, along with other key ... Continue Reading
Jon Toigo looks at the factors that determine whether disk or tape is more energy-efficient. Continue Reading
Analyst Jon Toigo explains why performing analytics on big data sets means that data is no longer considered cold, reducing the importance of ... Continue Reading