Fotolia
How do I enable file sharing on Ceph object storage?
The open source Ceph project prioritizes efficient data storage over easy access, so companies that want to set up file sharing on their object stores must do a little extra work.
Ceph object storage offers a fast way to store data, but setting up file sharing takes some work.
Under the hood, Ceph object storage consists of many storage nodes that chop files into binary objects and distribute them over object storage devices (OSDs). A typical Ceph configuration has hundreds or even more than a thousand OSD nodes.
Ceph offers four different ways users can access the object store:
- A Reliable Autonomic Distributed Object Store (RADOS) Gateway that's compatible with Amazon Simple Storage Service (S3);
- An iSCSI interface, which turns Ceph into an iSCSI SAN;
- CephFS, a POSIX-compliant file system that is locally mounted on the Ceph client; and
- The RADOS block device, which presents a typical Linux block device as if it were a local hard disk.
What about file sharing?
Providing file-sharing interfaces to Ceph is possible, but it must be through an additional layer to the object store. Companies can turn to the network file system, which allows administrators to create shares that give other Linux and Unix machines access to the NFS shares.
To provide NFS access to Ceph object storage, administrators must consider which Ceph access method to use. The S3-compatible interface is not a suitable candidate because it is mainly for cloud applications that must access storage directly using REST.
The RBD driver and the CephFS interface can provide NFS access to Ceph object storage, however. From the perspective of the Linux operating system, the RBD is just another block device on which an administrator could create any Linux file system, and the CephFS file system is just another POSIX-compliant file system, both of which can be shared using NFS shares.
