Posted by: Anonymous Coward
on December 06, 2004 01:43 AM
We already have an excellent next-generation network file system protocol: WebDAV. Properly implemented, it's been shown to be more efficient than SMB and NFS, there are lots of servers, and it has all the features (locking, versioning, metadata, security, authentication, etc.) you would want.
It doesn't work well in either Linux or Windows kernels, but that's not a problem with WebDAV, it's a problem with the kernels: any wide area networked file system is going to have lots of failures and the Linux and Windows kernels don't deal well with failures in their network file system code. They also don't have the hooks for dealing with something as general and powerful as WebDAV.
The real problem is that the kernel hackers have been driving network file systems, as opposed to people like the WebDAV developers. The kernel needs to become more general and robust, and the kernel network file system support needs to get better before anything is going to improve.
it's already here
Posted by: Anonymous Coward on December 06, 2004 01:43 AMIt doesn't work well in either Linux or Windows kernels, but that's not a problem with WebDAV, it's a problem with the kernels: any wide area networked file system is going to have lots of failures and the Linux and Windows kernels don't deal well with failures in their network file system code. They also don't have the hooks for dealing with something as general and powerful as WebDAV.
The real problem is that the kernel hackers have been driving network file systems, as opposed to people like the WebDAV developers. The kernel needs to become more general and robust, and the kernel network file system support needs to get better before anything is going to improve.
#