in one line: block storage vs file storage vs object storage
I was really confused until a colleague explained it to me!
So here it is, in extremely simple terms:
Block Storage: Application/user reads/writes blocks, constructs file itself. E.g. normal hard drive or SAN storage.
File Storage: Application/user reads/writes a particular file path in its entirety (typically using SMB/NFS/FTP/HTTP protocols) without worrying about the filesystem where it is stored. E.g. A block storage mounted as NFS on a remote server is a file storage for that remote server.
Object Storage: Application/user reads/writes a particular file without worrying about where it is stored in the directory hierarchy. The access is always made to the basename of the file in question. The object storage has to locate the file and provide access. Typically this is done by storing the SHA of file as well. The file is always located using this SHA value.
This also implies that any change in the file contents results in a new SHA value and hence the object storage treats as storing a new file and deleting the old one with the same name.
Don't forget that ultimately everything is stored on block storage only. The above three types of storages refer to the way the stored files are accessed rather than the way they are stored. Of course the block storage itself can be optimized and enhanced by vendors in a way that it serves as a better file storage or object storage.
Hope this helps!
I work for DellEMC Isilon: Word's No 1 file storage.
Know more about it here: https://www.dellemc.com/en-us/storage/isilon/index.htm
Cheers!! ๐
So here it is, in extremely simple terms:
Block Storage: Application/user reads/writes blocks, constructs file itself. E.g. normal hard drive or SAN storage.
File Storage: Application/user reads/writes a particular file path in its entirety (typically using SMB/NFS/FTP/HTTP protocols) without worrying about the filesystem where it is stored. E.g. A block storage mounted as NFS on a remote server is a file storage for that remote server.
Object Storage: Application/user reads/writes a particular file without worrying about where it is stored in the directory hierarchy. The access is always made to the basename of the file in question. The object storage has to locate the file and provide access. Typically this is done by storing the SHA of file as well. The file is always located using this SHA value.
This also implies that any change in the file contents results in a new SHA value and hence the object storage treats as storing a new file and deleting the old one with the same name.
Don't forget that ultimately everything is stored on block storage only. The above three types of storages refer to the way the stored files are accessed rather than the way they are stored. Of course the block storage itself can be optimized and enhanced by vendors in a way that it serves as a better file storage or object storage.
Hope this helps!
I work for DellEMC Isilon: Word's No 1 file storage.
Know more about it here: https://www.dellemc.com/en-us/storage/isilon/index.htm
Cheers!! ๐
Comments
Post a Comment