Distributed File System: Information and Heads up

Over the past couple of days, I have been doing some research and testing with the Distributed File System services of Windows 2003/2008.

I wanted to do a little write up about it, so that everyone who deals with user data will be prepared for this change.

Though this is a big change on how files will be accessible, the benefits of it will be much greater than the changes required.

I really encourage you to read the following write up on what DFS is: http://www.maxi-pedia.com/what+is+DFS

Once implemented DFS will do the following:

It will centralize the access method to files, so nothing will be linked to an actual physical server any more, but rather a logical representation of the data on it. So, the user, no matter what location they are at, or what server their data resides on, they would all access their data by navigating a hierarchy of folders start with \\chino.k12.ca.us\groupstore$\departments\business\
\\chino.k12.ca.us\groupstore$\departements\humanresources\  etc …

These folders can be on different physical servers, but the logical pointers are all the same. This will make it much easier to find data, and all the heavy lifting will happen in the DFS system.

You may be wondering why I’m pushing to move this direction. The above article does describe the benefits of DFS, but here are the main points:

1- With DFS, we can have fault tolerance, and fail over in case a server goes down. The data is effectively replicated to another server/storage device, and the users would seamlessly fail over to an active replica. In simple words: much more reliable data access.

2- Having a replica and failover, also means that the data is duplicated, which means, that we would now have a synchronized copy (a.k.a backup) of the data, in case of a loss of a server.

3- Because everything is replicated to two locations, and because there is a logical namespace to a physical server, we would now be able to perform data migrations for users during business hours without any impact on our users. (and will allow me to get a couple more hours of sleep at night)

4- The management of such a system will all be Active Directory integrated, and supported with Windows File Server Resource Manager (FSRM), which will allow us to establish quotas, and file type on user shared folders, therefore controlling what they save on our servers.

This will probably be a chance for the helpdesk for sure, and aside from adding basic folders for users, the management of the bulk of the DFS will be more of a sys admin job than the helpdesk, but if working properly, it’s a tried and true system, which has been in existence since Windows 2000 server.

Since I am currently working on moving all district data from do-fs2 to the SAN (Compellent), it makes sense that a DFS root would be established for this data, and attempt data migration that way.

Please do not panic with any of this, as I will be informing the necessary people with any changes they may encounter, or any learning they need to do in that topic. Until, I will keep you informed of any changes I make, and pending everything going as planned and researched, it should 100% transparent to the users.

If the first step of the DO data works well, the plan is to establish a DFS root for all the schools, and start having replicated data from the schools to the Compellent SAN, providing, backup, redundancy and fault tolerance.

For any questions, please talk to me.