For many decades now storage systems have come in essentially two flavors. File storage and Block storage. But now object storage is fast gaining popularity. Sometimes there is confusion as to what object storage offers that file and block do not. We all understand File and Block storage. We have been using it for many years now.
File storage is used to store files such as office documents, PDFs and other types of user files. File storage uses File Systems to store files by organizing them in a way that they can be located by specifying its path. A file has metadata associated with it such as owner who has access to the file and what type of access such as read, write, etc. Files are typically stored on a NAS (Network Attached Storage) system running on a local IP Network. NAS allows secure sharing of large number of files. NAS systems can scale to handle hundreds, thousands and even millions of files, but cannot scale infinitely.
We are also well versed with block storage which is typically used with database and OLTP application environments because these demand high performance. A block is simply a chunk of data. 4K blocks are common. A block has an address associated with it and it is accessed by sending a SCSI command to the storage system. Unlike a File which is organized and managed by a File System on a NAS storage, a block is under the direct control of the application. This results in better performance. A block has no meta-data associated with it such as the owner of the file, permissions, etc. A block is only meaningful in the context of the usage by the application which combines the blocks in a way that it requires to perform various functions.
Object storage systems allow relatively inexpensive, scalable and self-healing retention of massive amounts of unstructured data. Object storage is used for diverse purposes such as storing photos on Facebook, songs on Spotify, or files in online collaboration services, such as Dropbox.
Object storage is data (think of it as a file) with its associated metadata combined into an object. The object has an address calculated by combining the data and the metadata, known as the object ID. An object is retrieved by providing its object ID. Objects are stored in a flat file structure (as opposed to a hierarchical structure in a file system). Also, unlike file storage and NAS, objects can be distributed geographically. An object is not limited to any type or size of metadata. You can associate multiple types of metadata to describe the data including the name of the application that generated it, the level of protection used for the object and much more. This enables users to perform extensive analytics on objects because of the rich metadata that objects enable you to associate with them.
Object storage is ideal if you want a simple way to manage storage and offer a service that is spread across geographies and provides rich metadata that is user definable. Most object storage uses clusters of commodity servers with internal direct attached storage. HyperConverged systems for instance are well suited for object storage because you can scale storage by simply adding another node in a scale-out architecture. Object storage is popularly used by WebScale environments and Cloud Service providers such as Amazon and Google. DropBox for instance uses Object storage and it runs on Amazon S3 in the backend.
Data protection of object storage is handled a bit differently than file and block storage which primarily uses RAID based protection. Protection on object storage is either accomplished by replicating objects to one or more nodes or through a method such as erasure coding, which is essentially a RAID-like method that spans across geographies. Another unique for block storage is that it can be accessed via HTTP based REST APIs using simple commands such as Get, Put, Delete, etc.
So if you want storage that is easy to manage, scalable without limits, and which stretches across geographic boundaries in a single namespace, Object Storage is for you. Object storage is also a great way to move stale, unstructured data that is sitting on an expensive tier of SAN or NAS storage into a more affordable, albeit slightly lower performance object storage. It simply makes sense.
About the author:
Jaime is Marketing Manager at ISIT. He has worked in IT for over 25 years in various technical and non-technical roles in companies based in the US and the UAE. He has been a key contributor to ISIT since the company’s inception in 2007 in Sales, Business Development and Marketing. He is an avid follower of technology trends and enjoys talking about technology in a way that is easy for a non-technical person to comprehend.