Object Storage
Object storage is a computer data storage architecture that manages data as objects, as opposed to other storage architectures like file storage which manages data as a file hierarchy, and block storage which manages data as blocks within sectors and tracks.
Note: "Object storage is an alternative to NAS for handling unstructured data. There is speculation that object storage gradually will overtake scale-out NAS, but it's also possible the two technologies will continue to survive side by side. Both storage methodologies deal with scale, only in different ways. Object storage surfaced as a new method for easily scalable storage in web-scale environments. It often encompasses unstructured data that is not easily compressible, particularly large video files."
Examples...
Block: NVMe, NVMe-oF, iSCSI, Fibre Channel
Parallel file: Lustre, Spectrum Scale, Panasas, WekaIO, BeeGFS
Object: S3 (RESTful API)
Block Storage (SAN / "Structured")
Virtual servers, databases, etc...
Best suited for transactional data and frequently changing data.
Protocols: SCSI, Fibre Channel, SATA
Low Latency / High Performance.
Connected via FC and/or iSCSI.
Can be very expensive.
Limited Scalability.
Difficult to extend past the data center.
File Storage (NAS / "Unstructured")
Higher Latency / Higher Throughput.
Best suited for shared file data.
Protocols: CIFS and NFS
Connected via SMB, NFS, etc... and accessed via LAN.
Not as expensive as block.
Limited Scale Out.
Complex to manage at scale.
Difficult to extend past the data center.
Object Storage ("Unstructured")
Best for accessibility and reliability at scale. That scalability is the direct result of object storage's shared-nothing architecture and flat structure.
Storage backbone of the cloud.
Great for large sets of unstructured data.
Best suited for relatively static file data and as cloud storage.
Protocols: REST and SOAP over HTTP. The defacto standard RESTful API is currently S3
Ability to connect any type of device from anywhere.
Supports protocols like HTTP.
Parallelization becomes a key characteristic.
Object storage leverages metadata. There is no hierarchy to scan or crawl.
Millions of devices can access the same information simultaneously.
NOTE: Erasure coding is great for data resilience, but it adds significant latency for reads and writes. Erasure coding also doesn't protect against data corruption, human errors or malware/ransomware. Object storage, in general, doesn't provide snapshots, a staple for block and file storage that does provide that level of protection.
Object Storage examples: AWS S3, Azure Blob, Google Storage, Wasabi, MinIO
AWS Object Storage – Multiple performance and price points, including standard, auto-tiering, infrequent access, and archive.
Azure Object Storage – Premium, hot, cool, and archive performance tiers, each with pay-as-you-go or reserved capacity purchase options. Volume discounts apply across all objects.
Google Object Storage – Four performance tiers – Standard, Nearline, Coldline, and Archive