What kind of workload are you using Ceph for? (Part-2)

In part2 let’s look at CephFS, the second most popular interface being used with Ceph. CephFS provide a file system in a distributed manner so clients can read and write with the most common data type “File” directly without a single point of failure and performance could be scaled horizontally by adding more gateway and OSD. In order to provide the file service, metadata server MDS is required to support the extra bit of file structure and metadata. So the filesystem interface and metadata could be served over the object storage. After the Ceph version Pacific release, CephFS also provides multiple root filesystems, so it is possible to create a different repo for different usage. With ceph-fuse, you can mount CephFS basically on any OS and natively of course it supports Linux and as well Windows. Besides mounting the CephFS directly, the client can also receive files service on top of NFS or Samba.

CephFS normally is being consumed directly by an OS, it will be mounted natively or using FUSE to allow applications running on top of it. It makes sense of using CephFS because of the limitation of what a traditional FS couldn’t provide, e.g. horizontal scaling.

Using a Linux VM should be fairly straightforward, having a native driver to mount CephFS directly, or at least CephFS-FUSE should do the job. In the Windows environment…