When making technical comparisons in IT, using an analogy is often helpful. IBM did a presentation on object storage vs file storage, recently. The presenter’s analogy on these two types of data storage was this:
File Storage
File storage is like parking your car in a parking lot. You park your vehicle in a space which enables you to access your car directly, within the parking spot, any time you need to. If you are shopping and you want to drop off packages, you simply return to your vehicle, open the trunk, deposit your purchases, and return for more shopping. You are the only one who knows where your car is parked, which is how file storage works. With file storage, you can save your file (unstructured data) in a folder, using a name you’ve created, and you can access and edit the file directly through the operating system by referencing that given name.
Object Storage
In contrast to self-parking your car at the mall, object storage is tantamount to valet parking. You simply pull up, leave your keys in the car, and get a ticket so your vehicle can be retrieved by someone else. Any valet worker can retrieve your car because the ticket has the meta data as to where your car is located. You cannot access your car directly, like you can with self-parking, but you can submit a request to retrieve your car when you need it back. The time it takes to get your car back may take longer than self-parking, due to the wait for a valet worker to retrieve your car, but the performance is still acceptable to leave a tip, in most cases.
The valet ticket is the unique identifier and corresponds with the make, model and location of the car, which can be thought of as the object ID. Only applications understand object IDs, and that is why object storage is application specific, unlike a file system that can be accessed by anyone with the correct permissions. An object within object storage is immutable because you cannot modify it in place. Objects are written into object storage with the PUT command and retrieved with the GET command. The GET command enables you to read objects and start their transfer of data. Contrary to naming and storing your file in a file storage environment, object storage creates its own object ID and does not use file naming conventions.
Use Cases for File Storage and Object Storage
With file storage, you can save your file in a directory of your choice, and you can access the file directly through the operating system to make edits at any time. This is often referred to as user shares. Think of creating a document in MS Word, for example, and storing it on your personal drive or perhaps on a file share such as Google Drive or MS Sharepoint.
A popular use case for object storage is data protection (backup and archive applications). A characteristic of object storage is that it is highly scalable and resilient, which is a requirement for data protection storage. Moreover, it often costs less than file, or block storage. IBM’s Cloud Object Storage, ICOS, formerly known as Cleversafe, is a leader in object storage deployments for backup applications like Spectrum Protect, Netbackup and CommVault, because it can replace both virtual and physical tape systems.
Objects like images or video files are good candidates for object storage because they are larger in size, and not opened to be modified, most of the time. Because of their storage consumption, and inability to be deduped or compressed, the lower cost, lower performing characteristics of object storage makes them an ideal platform for these media types. Issuing a GET command to start a video feed, or image retrieval, is where large content stores like Shutterfly and Apple iTunes have used object storage successfully.
File Storage Gateways
Using file storage gateways, in front of object storage devices, extends object storage use cases. The gateways present the “D:\” so to speak… but it uses object storage on the back end. They provide the translation, so a file can be stored in object storage. There are several File Storage Gateway technologies available. As a follow-on to this blog, I will discuss gateways in more detail.
Conclusion
So, there you have the key differences of file storage to object storage. Simply put, file storage is to self-parking as object storage is to valet-parking. In either case, they both allow you to store and retrieve your data, albeit with different characteristics, which in turn, helps determine what data sets should be stored in each storage medium. User shares is a great fit for file storage, so users can access their data in a self-service format. Conversely, object storage is perfect for streaming videos because it requires a valet service like Netflix to retrieve the videos for you.
Mainline offers storage solutions across the entire spectrum of technologies to suit your business needs and budget. Contact your Mainline Account Executive, or contact Mainline directly for more information on storage systems for your data center, the cloud, or remote sites.
Additional Information:
Difference between Software Defined Storage and Storage Virtualization