三种存储的差别，摘抄openstack文章

yxxs123

　　具体区别可以看看这 ~~http://blog.sina.com.cn/s/blog_6795e96f010130e2.html
　　Object storage


In OpenStack: Object Storage service (Swift)


Related concepts: Amazon S3, Rackspace Cloud Files, Ceph Storage


With object storage, files are exposed through an HTTP interface, typically with a REST API. All client data access is done at the user level: the operating system is unaware of the presence of the remote storage system. In OpenStack, the Object Storage serviceprovides this type of functionality. Users access and modify files by making HTTP requests. Because the data access interface provided by an object storage system is at a low level of abstraction, people often build on top of object storage to build file-basedapplications that provide a higher level of abstraction. For example, the OpenStack Image service can be configured to use the Object Storage service as a backend. Another use for object storage solutions is as a content delivery network (CDN) for hostingstatic web content (e.g., images, and media files), since object storage already provides an HTTP interface.


Block storage (SAN)


In OpenStack: Block Storage service (Cinder)


Related concepts: Amazon Elastic Block Store (EBS), Ceph RADOS Block Device (RBD), iSCSI


With block storage, files are exposed through a low-level computer bus interface such as SCSI or ATA, that is accessible over the network. Block storage is synonymous with SAN (storage area network). Clients access data through the operating system at the devicelevel: users access the data by mounting the remote device in a similar manner to how they would mount a local, physical disk (e.g., using the "mount" command in Linux). In OpenStack, the cinder-volume service that forms part of the Compute service providesthis type of functionality, and uses iSCSI to expose remote data as a SCSI disk that is attached to the network.


Because the data is exposed as a physical device, the end-user is responsible for creating partitions and formatting the exposed disk device. In addition, in OpenStack Compute a device can only be attached to one server at a time, so block storage cannot beused to share data across virtual machine instances concurrently.


File storage (NAS)


In OpenStack: none


Related concepts: NFS, Samba/CIFS, GlusterFS, Dropbox, Google Drive


With file storage, files are exposed through a distributed file system protocol. Filesystem storage is synonymous with NAS (network attached storage). Clients access data through the operating system at the file system level: users access the data by mountinga remote file system. Examples of file storage include NFS and GlusterFS. The operating system needs to have the appropriate client software installed to be able to access the remote file system.


Currently, OpenStack Compute does not have any native support for this type of file storage inside of an instance. However, there is a Gluster storage connector for OpenStack that enables the use of the GlusterFS file system as a back-end for the Image service.

　　

　　Object storage

In OpenStack: Object Storage service (Swift)

Related concepts: Amazon S3, Rackspace Cloud Files, Ceph Storage

With object storage, files are exposed through an HTTP interface, typically with a REST API. All client data access is done at the user level: the operating system is unaware of the presence of the remote storage system. In OpenStack, the Object Storage serviceprovides this type of functionality. Users access and modify files by making HTTP requests. Because the data access interface provided by an object storage system is at a low level of abstraction, people often build on top of object storage to build file-basedapplications that provide a higher level of abstraction. For example, the OpenStack Image service can be configured to use the Object Storage service as a backend. Another use for object storage solutions is as a content delivery network (CDN) for hostingstatic web content (e.g., images, and media files), since object storage already provides an HTTP interface.

Block storage (SAN)

In OpenStack: Block Storage service (Cinder)

Related concepts: Amazon Elastic Block Store (EBS), Ceph RADOS Block Device (RBD), iSCSI

With block storage, files are exposed through a low-level computer bus interface such as SCSI or ATA, that is accessible over the network. Block storage is synonymous with SAN (storage area network). Clients access data through the operating system at the devicelevel: users access the data by mounting the remote device in a similar manner to how they would mount a local, physical disk (e.g., using the "mount" command in Linux). In OpenStack, the cinder-volume service that forms part of the Compute service providesthis type of functionality, and uses iSCSI to expose remote data as a SCSI disk that is attached to the network.


Because the data is exposed as a physical device, the end-user is responsible for creating partitions and formatting the exposed disk device. In addition, in OpenStack Compute a device can only be attached to one server at a time, so block storage cannot beused to share data across virtual machine instances concurrently.

File storage (NAS)

In OpenStack: none

Related concepts: NFS, Samba/CIFS, GlusterFS, Dropbox, Google Drive

With file storage, files are exposed through a distributed file system protocol. Filesystem storage is synonymous with NAS (network attached storage). Clients access data through the operating system at the file system level: users access the data by mountinga remote file system. Examples of file storage include NFS and GlusterFS. The operating system needs to have the appropriate client software installed to be able to access the remote file system.

Currently, OpenStack Compute does not have any native support for this type of file storage inside of an instance. However, there is a Gluster storage connector for OpenStack that enables the use of the GlusterFS file system as a back-end for the Image service.