DFS or Sharepoint
Hi
i have some teams in our company that need access to the same set of data. Problem is that they are in two different geographical locations connected via slow wan links.
Now i had thought of implementing DFS Namespaces together with DFS Replication and i think that would work ok for me. But some questions arose from that ie is that the best way to do it ie how bandwidth hungry is it, file locking etc.
The other scenario i had thought of was Sharepoint using doc libraries now i know this woul dcure my file locking problem right away but access to the data from the second geographical location would be quite slow as it would have to traverse the slow wan link in order to open a file whereas with DFS each site has a local copy of the data and the changes are all thats synced up the wan links.
i am right in what say or totally wrong?
cheers
i have some teams in our company that need access to the same set of data. Problem is that they are in two different geographical locations connected via slow wan links.
Now i had thought of implementing DFS Namespaces together with DFS Replication and i think that would work ok for me. But some questions arose from that ie is that the best way to do it ie how bandwidth hungry is it, file locking etc.
The other scenario i had thought of was Sharepoint using doc libraries now i know this woul dcure my file locking problem right away but access to the data from the second geographical location would be quite slow as it would have to traverse the slow wan link in order to open a file whereas with DFS each site has a local copy of the data and the changes are all thats synced up the wan links.
i am right in what say or totally wrong?
cheers
Andres Perales
Does your remote site have a connection to the internet, is that connection faster than your WAN Link, if so I would use SharePoint.
ASKER
well its the same speed, how will that help?
We currently are using both DFS and Sharepoint for document management. In my opinion I would probably go with sharepoint for the situation you have described. Depending on the amount of people that are accessing the files you could run into some serious issues with DFS and last modified files. Basically this is the file locking thing you mentioned.
You could have two people working on the same document on each end of the DFS ring and since they are both working on individual files on two servers one of them is going to lose the changes they made to the files. I have had this happen and the users are definitely not happy when that happens. The more people you have accessing the files the higher the chance this will happen.
DFS can also give you some issues when attempting to back it up. We were never really able to get a good backup using the Volume shadow copy method that MS reccomends. We ultimately had to add pre and post commands to stop the DFS service then back up the data then start the DFS service again.
You could have two people working on the same document on each end of the DFS ring and since they are both working on individual files on two servers one of them is going to lose the changes they made to the files. I have had this happen and the users are definitely not happy when that happens. The more people you have accessing the files the higher the chance this will happen.
DFS can also give you some issues when attempting to back it up. We were never really able to get a good backup using the Volume shadow copy method that MS reccomends. We ultimately had to add pre and post commands to stop the DFS service then back up the data then start the DFS service again.
I HIGHLY recomend using SharePoint Services 3.0
http://technet.microsoft.com/en-us/windowsserver/sharepoint/bb400747.aspx
It will do what you need to do plus much much much more!
http://technet.microsoft.com/en-us/windowsserver/sharepoint/bb400747.aspx
It will do what you need to do plus much much much more!
ASKER
but can someone explain how it will give local speed to users at both locations the way dfs can by replicating the changes of two local folders.
i need to be able to ahve the both locations access the files and folders locally and not on e site access over the wan link.
i need to be able to ahve the both locations access the files and folders locally and not on e site access over the wan link.
ASKER
also what size can the sharepoint doc library grow to, we have some very large foders
Microsoft discourages growth beyond 100GB
http://technet.microsoft.com/en-us/library/cc262787.aspx
To answer your other question, users will be able to "check out" the documents to work on them & then "check in" when finished. Users will be able to open & view the docs at any time but will be notified it is checked out for editing & by whom. If a user checks it out, it will be saved locally until it is checked back in.
http://technet.microsoft.com/en-us/library/cc262787.aspx
To answer your other question, users will be able to "check out" the documents to work on them & then "check in" when finished. Users will be able to open & view the docs at any time but will be notified it is checked out for editing & by whom. If a user checks it out, it will be saved locally until it is checked back in.
If you are looking for a solution that is more along the lines of replicating a whole file server structure then you will probably need to use DFS. If you are looking for a solution that will allow you to post a small amount (less than the 100gb wantabe mentioned) then sharepoint may work better for you.
However in your question you mentioned that you do not want the users to have to pull data over the WAN link. If that is the case then you pretty much have to use DFS so that you can have a local copy of the files in each location.
What exactly is your WAN link connection?
I was running a remote office with about 200gb worth of user data in a DFS hub and spoke over a T1 line. It took a long time for the initial sync since I didnt pre-seed the data but after that the replication kept the files up to date pretty well. One suggestion I can make especially over a slow link is that you will want to increase the size of your staging folder well beyond the default of 4gb. The larger the staging area the more efficient your RDC will be and the less amount of data you will need to transfer over the WAN link
However in your question you mentioned that you do not want the users to have to pull data over the WAN link. If that is the case then you pretty much have to use DFS so that you can have a local copy of the files in each location.
What exactly is your WAN link connection?
I was running a remote office with about 200gb worth of user data in a DFS hub and spoke over a T1 line. It took a long time for the initial sync since I didnt pre-seed the data but after that the replication kept the files up to date pretty well. One suggestion I can make especially over a slow link is that you will want to increase the size of your staging folder well beyond the default of 4gb. The larger the staging area the more efficient your RDC will be and the less amount of data you will need to transfer over the WAN link
ASKER
xxdcmast: that sounds like an excellent expierenced answer, what did yo end up using as the staging folder size?
can i send a copy of the data to the second site on exteranl hdd and then use some dfs tools so the inital replication is kept to a minimum, if so what are the steps for this to work?
cheers
can i send a copy of the data to the second site on exteranl hdd and then use some dfs tools so the inital replication is kept to a minimum, if so what are the steps for this to work?
cheers
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
xxdcmast following on from the pre seeding comment, i have simply copied one folder from one server to another and know i would like to have a replication group setup within dfs to keep this folder up to date. do i simply point the replication group at the folder i copied to server2 or do i need to place that copied folder somewhere special like prestaging fodlers or anything?
If you just did a regular windows copy I think you will run into issues. The way DFS does the comparisons is based on file hashes. A regular copy option will not keep these hashes the same.
All you will need to do is place the copied folder structure in any location you want on the server. As long as you target both of the folders on each server at the same level in the folder structure you should be all set. You dont have to move the folders into pre-existing or anything
All you will need to do is place the copied folder structure in any location you want on the server. As long as you target both of the folders on each server at the same level in the folder structure you should be all set. You dont have to move the folders into pre-existing or anything
ASKER
so how should i copy? robocopy or ntbackup or soemthing like that you reckon?
cheers
cheers
If you look in this link he details several options including ntbackup, xcopy, and robocopy along with their proper switches.
http://blogs.technet.com/askds/archive/2008/02/12/get-out-and-push-getting-the-most-out-of-dfsr-pre-staging.aspx
I used robocopy myself and used the switches he reccomended to me in the comments section.
http://blogs.technet.com/askds/archive/2008/02/12/get-out-and-push-getting-the-most-out-of-dfsr-pre-staging.aspx
I used robocopy myself and used the switches he reccomended to me in the comments section.
ASKER
cheers much appreciated
ASKER
one last point maybe you can help with.
one of the folders i am replciating is only 855MB however if i check the folder size on the server i replicated to it says tis 4.5gb and if i check it, inside the dfsprivate folder there is a staging folder that is 4GB in size, what is that and can i ever delete the contents?
cheers
one of the folders i am replciating is only 855MB however if i check the folder size on the server i replicated to it says tis 4.5gb and if i check it, inside the dfsprivate folder there is a staging folder that is 4GB in size, what is that and can i ever delete the contents?
cheers
No you dont want to delete the contents of your staging folder. The fact that it is 4gb makes sense and it means that DFS is working properly. Before a file is replicated it is brought into the staging folder and checked for changes and then only the changes are replicated across the WAN.
The DFS service will automatically do cleanups on the staging folder and log an event in your event logs when it does. So keep an eye on the event logs and you will see events speaking about the staging high water level and the automatic cleanup generated when this level is hit.
The DFS service will automatically do cleanups on the staging folder and log an event in your event logs when it does. So keep an eye on the event logs and you will see events speaking about the staging high water level and the automatic cleanup generated when this level is hit.
ASKER
ahh so thats whay you increased your staging folder the more changes you suspect will happen inside that folder then hence the need for a bigger staging folder.
pretty neat this dfs stuff could be really useful for me.
do you come across many problems with it?
pretty neat this dfs stuff could be really useful for me.
do you come across many problems with it?
The idea behind the larger staging folder is that the more stuff you can hold in the staging folder on each end the higher the chance there is already a document that has been compared and hashed so the less you put across the wire.
I have had some issues with DFS.
1. The file locking and file overwrite issue I mentioned above. Last one to save the file wins. So you could have someone work on a file for an hour save it and close but at the same time someone else makes 1 small change and saves it. If the small change person saves it last that whole hour of work will be lost. You may be able to find the file in the conflict and deleted folder but I have no had great luck with that.
2. Backups using the MS volume shadow copy reccomendation were running incredibly slow for us using backup exec 12.5. We could not get this to work correctly so we have to issue pre and post commands to stop and start the DFS so we can run our backups.
3. You need to make sure that you virus scan program does not modify the file date and times when it scans otherwise you can have serious issues with files needing to be re-replicated.
I have had some issues with DFS.
1. The file locking and file overwrite issue I mentioned above. Last one to save the file wins. So you could have someone work on a file for an hour save it and close but at the same time someone else makes 1 small change and saves it. If the small change person saves it last that whole hour of work will be lost. You may be able to find the file in the conflict and deleted folder but I have no had great luck with that.
2. Backups using the MS volume shadow copy reccomendation were running incredibly slow for us using backup exec 12.5. We could not get this to work correctly so we have to issue pre and post commands to stop and start the DFS so we can run our backups.
3. You need to make sure that you virus scan program does not modify the file date and times when it scans otherwise you can have serious issues with files needing to be re-replicated.