I've been googling for ages trying to find information about what object storage is in relation to cloud computing but have been unable to find anything. Can anybody provide any links with succinct overviews of it?
Its not the concept of cloud storage in general I'm looking for information on, its specifically object storage, what is object storage and how does it differ from other types of storage.
Source http://www.softlayer.com/services/storagelayer/object-storage
"What is object-based storage?"
"It’s easiest to understand object-based storage by comparing it with other types of remote storage-block and file. A block-level storage system (such as iSCSI or most SANs) is, essentially, raw storage with no file system of its own—it leaves file system concerns to the client. In contrast, file-level storage devices (such as a NAS device) have a file system and protocol (such as NFS and SMB/CIFS) for accessing and managing data. An object-based storage system creates data containers—objects—and then writes fixed-sized blocks of un-interpreted bytes within the container, as well as assigned metadata for the object itself."
I was here looking for some information, found this question.... figured I'd post this response as an FYI, as I think the questioner was asking about the same thing as I was looking for.
At a recent SNC conference, there was a panel discussion on this very topic of object storage but not specifically in relation to cloud computing. Object storage, as one member of the panel explained, is akin to obtaining a key when you park your car inside the garage. The same key can be used to retrieve the car from the garage. So object storage is a way to attach metadata to the content stored. It becomes easy to retrieve it. The larger point is about the manageability of storage area networks. New applications are emerging in the marketplace for retrieval and aggregation of objects of storage including the copies mirrored in other sites. These applications can also help to retrieve corrupted data or data lost due to disasters.
From the link you posted, this seems to be referring to a technology that is similar to what Gigaspaces (commercially) and JavaSpaces (FOSS) has been doing for some time. It is technology that takes 'live' objects (instances of classes) and spreads them out amongst runtimes on multiple machines, then abstracts access to them via a common API. These solutions often allow you to run code 'close to the data', usually serialize objects to disk when space gets tight, and take care of surviving individual machine failure.
Storage Types:- Block storage, File storage, Object storage.
From my understanding, Object storage falls in-between the Block and File storage types. Why? The level of abstraction (storage interface) in the Object storage type is good enough for faster access (like the Block storage type - low level) and secure data sharing (like the File storage type - high level).
To explain in simple terms, retrieving blocks is faster but this compromises security part of data sharing since there is no metadata for a block of data (unlike a file which can maintain metadata).
Objects are storage containers with a file-like interface to store any type of data, such as files, database records etc. A single object could even be used to store an entire file system or database.
For more details, http://www.storagevisions.com/White%20Papers/MesnierIEEE03.pdf
