| Author |
Message |
Dave
Guest
|
Posted:
Mon Oct 17, 2005 4:50 pm Post subject:
DFS for 100% uptime? |
|
|
Hi folks,
I'm thinking about implementing DFS for my network to increase uptime,
the idea would be something like this...
2 Servers on same domain, same site. Server1 and Server2
both servers have a D partition which is where the DFS roots will
reside.
I'll setup a root folder on "Server1 D:" and then a new root target on
"Server2 D:"
I'll then create folders with the same name on both servers in the root
of D:
folders from server1 will then be added as links under the DFS root on
Server1 and new link targets added using folders on Server2
On the clients I map drives to the linked folders using
Server1\dfsroot\foldername
This works ok. When one of the servers falls over I can still get at my
data.
I have a couple of questions though regarding this setup..
is it correct?
is there a better way?
I'd like to have all clients update files on only server1, then when it
fails the clients use server2, then fall back to server1 when it comes
back online, is this possible with current setup or will I have to put
server2 into another site and create costs etc?
how do I restore DFS in case one of the servers fail, I wouldn't want
to rebuild a new DFS on a newly built server then the blank directories
overwrite the existing stuff on the other server resulting in all data
being lost?
What does the topology represent on the replication side of things, hub
and spoke, mesh etc.?
any help is appreciated
Dave |
|
| Back to top |
|
 |
Dave
Guest
|
Posted:
Tue Oct 18, 2005 8:51 am Post subject:
Re: DFS for 100% uptime? |
|
|
| anybody? |
|
| Back to top |
|
 |
III
Guest
|
Posted:
Sat Oct 22, 2005 8:50 pm Post subject:
RE: DFS for 100% uptime? |
|
|
DFS is great when you want to share or replicate lots of static files. It's
not so great when you have lots of files that change constantly. The
replication traffic is a killer because you send the whole file each time
rather than the changes. Microsoft is not there yet. Next release of Windows
promises to offer bit level replication so just to send change.
We started out in the scenario you outline but after the FRS stop working
because the replciation could not keep up. We switch to WAFS from
http://www.availl.com/
It works like a champ!
You could also try NSI Double-Take or Replistore.
"Dave" wrote:
| Quote: | Hi folks,
I'm thinking about implementing DFS for my network to increase uptime,
the idea would be something like this...
2 Servers on same domain, same site. Server1 and Server2
both servers have a D partition which is where the DFS roots will
reside.
I'll setup a root folder on "Server1 D:" and then a new root target on
"Server2 D:"
I'll then create folders with the same name on both servers in the root
of D:
folders from server1 will then be added as links under the DFS root on
Server1 and new link targets added using folders on Server2
On the clients I map drives to the linked folders using
Server1\dfsroot\foldername
This works ok. When one of the servers falls over I can still get at my
data.
I have a couple of questions though regarding this setup..
is it correct?
is there a better way?
I'd like to have all clients update files on only server1, then when it
fails the clients use server2, then fall back to server1 when it comes
back online, is this possible with current setup or will I have to put
server2 into another site and create costs etc?
how do I restore DFS in case one of the servers fail, I wouldn't want
to rebuild a new DFS on a newly built server then the blank directories
overwrite the existing stuff on the other server resulting in all data
being lost?
What does the topology represent on the replication side of things, hub
and spoke, mesh etc.?
any help is appreciated
Dave
|
|
|
| Back to top |
|
 |
Paul Williams [MVP]
Guest
|
Posted:
Sun Oct 23, 2005 4:50 pm Post subject:
Re: DFS for 100% uptime? |
|
|
As III states, this is very much doable, but it is very inefficient in
comparison to what we will get in R2. However, with 2003 you can do what
you want pretty well.
If I read your post correctly yes. The root, and the actual shares are
separate. And you will need to replicate the root for full redundancy
(right click on it and add a replica).
| Quote: | is there a better way?
|
Yes. But it will cost. In R2 this answer will be a flat no.
| Quote: | I'd like to have all clients update files on only server1, then when it
fails the clients use server2, then fall back to server1 when it comes
back online, is this possible with current setup or will I have to put
server2 into another site and create costs etc?
|
This is possible with some manual intervention. You can state that one of
the replicas isn't used for referrals. If there's a failure, you change
this (takes a second).
| Quote: | how do I restore DFS in case one of the servers fail, I wouldn't want to
rebuild a new DFS on a newly built server then the blank directories
overwrite the existing stuff on the other server resulting in all data
being lost?
|
From a replication stand-point, you would just do a non-authorative restore
of FRS. You would do this after restoring your server. This is just a
registry value and a restart of FRS. Search for Burflags and D2 (I've
mentioned this a lot in this group).
| Quote: | What does the topology represent on the replication side of things, hub
and spoke, mesh etc.?
|
This is based on your AD replication topology (or more specifically, your
site link objects). Intrasite is going to be mesh.
In R2 you can choose.
--
Paul Williams
Microsoft MVP - Windows Server - Directory Services
http://www.msresource.net | http://forums.msresource.net |
|
| Back to top |
|
 |
Dave
Guest
|
Posted:
Sun Oct 23, 2005 4:50 pm Post subject:
Re: DFS for 100% uptime? |
|
|
thanks guys,
you've helped a lot! I've checked out Availl's website and contacted
them about their product, it's very expensive so I'll probably not
bother. Due to the limitations of FRS I think I'll wait until R2 and
see what that's like.
cheers guys
Dave |
|
| Back to top |
|
 |
|
|
|
|