Distributed file system wisdom?
Apr. 18th, 2007 10:43 amNow that our main user block is all on Active Directory, I've been asked to look into Distributed File System. A bit of background: We have several Windows 2003 Cluster Servers each providing file shares (for illustration USERS, GROUPS and PRINT). These present friendly names to the punters and hold a big ol' pile of enterprise-scale goodness.
Now we're on AD, it would be really neat to bring everything together into one domain namespace. I would like to present users with the following:
It's super-clean, makes life easy for users and support staff as well, and completely insulates the users from the hardware. But I have a sneaking feeling that, since these are already cluster resources, we can't do this without presenting the cluster name (or some made-up fakesimile of it) in the name, eg, \\exe.nhs.us\users\fredbloggs$. That really has no benefit to us over the existing \\users\fredbloggs$ arrangement, and in fact has disbenefits, since the clusters auto-share their resources and the DFS resources don't so we'd have to go backwards to handmade (and therefore error-prone and slow-to-create) shares.
I'm thinking that in our situation, what I want is not possible, and what is possible is no better than what we have. Am I wrong?
Now we're on AD, it would be really neat to bring everything together into one domain namespace. I would like to present users with the following:
- \\exe.nhs.uk\fredbloggs$ (currently on \\USERS\fredbloggs$)
- \\exe.nhs.uk\cardiology$ (currently on \\GROUPS\cardiology$)
- \\exe.nhs.uk\printer1 (currently on \\PRINT\printer1)
It's super-clean, makes life easy for users and support staff as well, and completely insulates the users from the hardware. But I have a sneaking feeling that, since these are already cluster resources, we can't do this without presenting the cluster name (or some made-up fakesimile of it) in the name, eg, \\exe.nhs.us\users\fredbloggs$. That really has no benefit to us over the existing \\users\fredbloggs$ arrangement, and in fact has disbenefits, since the clusters auto-share their resources and the DFS resources don't so we'd have to go backwards to handmade (and therefore error-prone and slow-to-create) shares.
I'm thinking that in our situation, what I want is not possible, and what is possible is no better than what we have. Am I wrong?