One of the really handy things about Microsoft’s Distributed File System (DFS) is that it allows you to replicate files between file servers. This is handy if you have users working in multiple locations and want to provide local file access to keep them from having to download files across a WAN link. As useful as Windows Server DFS replication is, however, there are some things that you need to consider as you deploy it.
One of the first things to consider is that DFS replication was never intended as a replacement for traditional backups. Remember, if a file becomes corrupted, is incorrectly modified or is encrypted by ransomware, that modified file will be replicated. Replication does not give you a way of reverting the file back to its previous state. For that, you will need backup and recovery software.
Another thing to consider before enabling replication is the type of replication group that you want to create.
DFS replication is based on the use of replication groups, and there are two types of replication groups available.
- Data collection replication group: The first type is a replication group for data collection. This type of replication group is typically used for data consolidation, especially when there is a need to simplify data backups. Suppose for a moment that an organization has a main office and a branch office. A replication group that is designed for data collection could be established between the DFS servers at the two locations. In doing so, all of the file data from the DFS servers at the branch office would be replicated to the main office. That way, the backup software running in the main office would be able to create a backup of the data from the branch office because a copy of that data resides on DFS servers within the main office.
- Multipurpose replication group: The other option for creating a replication group is to create a multipurpose replication group. As its name implies, a multipurpose replication group can be used for almost any purpose.
If you’re in doubt as to which type of replication group to create, it’s usually best to go with a multipurpose replication group, as shown in Figure 1.
A multipurpose replication group is the most flexible option.
Another thing that you need to be thinking about before you create a replication group is where the replication group will reside within Active Directory. The replication group exists at the domain level, and it has to be assigned a name that is unique within the domain. You can see what the name and domain assignment screen looks like in Figure 2.
You will have to choose a name that is unique within the selected domain.
One of the biggest decisions that you will need to make before you begin implementing DFS replication is the replication topology that will be used to replicate data among the group members. DFS gives you three different options
- Hub and spoke topology: A hub and spoke topology requires at least three replica members. One of the replica members will act as a hub, and all others will act as spokes. This topology tends to be the most useful in situations in which data originates from a central source (the hub) and needs to be replicated out to other locations (the spokes).
- Full mesh replication topology: In this topology, each member of the replication group replicates data with every other member. Microsoft generally recommends that you use a full mesh only if you have 10 or fewer members in the replication group.
- Custom topology: Rather than having Windows force you into a generic replication topology, you can create your own custom replication topology outside of the New Replication Group Wizard. You can see what the three replication options look like in Figure 3.
This is what the topology selection screen looks like.
Before you begin replicating data within your Distributed File System, you will need to consider whether to use the full bandwidth available or to limit bandwidth by, for example, limiting the replication process so that it occurs only during off-peak periods. DFS is now less bandwidth-intensive than it once was because the replication process works at the block level rather than the file level, so only modified blocks are replicated. You can see what the Replication Group Schedule and Bandwidth screen looks like in Figure 4.
You can replicate data according to a preset schedule.
One last thing that you probably need to think about is which folders should be replicated. It’s easy to assume that all folders will need to be replicated to all of the replication group members. In reality, though, there may be folders that are used only in one specific location, so it may not make sense to replicate them. As you can see in Figure 5, DFS gives you the flexibility to choose which folders you want to replicate.
You don’t necessarily have to replicate every folder.