DFS Replication Fault Tolerance

[1UnixNewbie]

Hello,

I wanted to set up a DFS Root Replica, so if one of my 2 DC's went down, my users would not experience an outage. Of course unless they already had a file open, in which they would just have to close and reopen.

Problem:
I have one site, with both DC's on the same segment, and actually the same switch.
We have a sales database program that is currently using DFS along with other files that are accessed and modified all day long. Some people leave the files open all day.
It appears that FRS does not replicate data until the file has been closed, which causes my problem. The last file closed wins in replication, which would cause data inconsistancy pretty quickly.

Is there a way to enforce fault tolerance with no load balncing for DFS so my users can only use one DFS server, and leave the other one alone except for replication and fault tolerance????

Any suggestions welcome...

 

[svsawkar]

Nope. DFS was never designed for any type of fault tolerance- it's meant to extend a namespace for easy searching of files and folders.

FRS was also not designed for fault tolerance- it's meant to replicate relatively static data. Last writer wins is the method and file changes are not replicated- the entire file goes.

If you want fault tolerance, seek a 3rd party solution or look into clustering some file servers.

Or write your own 'poor man' version of file replication with robocopy and the scheduler.

Sorry!

/Siddharth