Understanding Windows 2000 DFS Replication Topologies

In Microsoft Windows 2000, Distributed File System (DFS) replication enables automated synchronization of shared folders across multiple servers to ensure high availability and load balancing. Implemented exclusively in domain-based DFS namespaces, this feature relies on the underlying File Replication Service (FRS) to automatically distribute updates among replica set members. Windows 2000 uses predefined and customizable logical network topologies—such as Ring, Hub and Spoke, and Full Mesh—to determine the precise communication pathways and schedules for synchronizing data across the enterprise.

The Role of File Replication Service (FRS)

Windows 2000 DFS does not replicate files on its own; it delegates the task to the File Replication Service (FRS). FRS is a multi-master replication engine integrated with Active Directory. When an administrator creates a replica set for a domain-based DFS root or link, FRS handles state synchronization across all configured targets.

FRS uses the NTFS Update Sequence Number (USN) change journal to detect modifications made to local files. Once a file or folder is modified, FRS stages the change and transmits it across logical connection objects defined by the chosen topology. Because it is multi-master, updates can occur on any replica partner and will propagate throughout the topology.

Built-In Topology Configurations

When configuring replication for a DFS link or root with three or more targets using the Distributed File System snap-in (dfsgui.msc), Windows 2000 prompts the administrator to select a replication topology. Windows 2000 provides several predefined models:

1. Ring Topology

In a ring topology, each replica server connects to two neighboring servers, forming a closed continuous loop.

  • Data Flow: Updates travel sequentially around the ring until all members are synchronized.
  • Redundancy: If a single link or server fails, replication can continue along the opposite direction of the loop.
  • Best Use: Suited for environments with geographically dispersed servers with similar bandwidth constraints.

2. Hub and Spoke Topology

The hub and spoke model designates one central server (the hub) and connects all secondary servers (the spokes) directly to it.

  • Data Flow: Spokes communicate solely with the hub. Spoke-to-spoke communication must route through the central server.
  • Redundancy: If the hub fails, replication halts between all spokes until the hub recovers.
  • Best Use: Ideal for branch-office scenarios where central headquarters houses the primary server and branch offices possess dedicated, individual connections back to corporate.

3. Full Mesh Topology

In a full mesh topology, every member server establishes a direct replication connection with every other server in the replica set.

  • Data Flow: Updates transmit directly from the source server to every other partner without intermediary hops.
  • Overhead: Connection counts scale quadratically with the formula \(N(N-1)/2\) where \(N\) is the number of servers.
  • Best Use: Best for small replica sets (typically under 10 servers) residing on high-speed local area networks (LANs).

4. Custom Topology

Administrators can define custom topologies manually by explicitly generating connection objects between specific pairs of servers. This approach is used when standard geometric topologies do not align with physical WAN link constraints or organizational traffic policies.

Active Directory Integration and Scheduling

Windows 2000 stores replication topologies as Active Directory objects within the domain partition under the CN=File Replication Service container.

Each directional link between two members is represented by an nTDSConnection object. Windows 2000 allows administrators to control network bandwidth consumption by configuring replication schedules on these connection objects:

  • Time Windows: Connections can be scheduled in 15-minute intervals to permit or deny replication during business hours.
  • Staggered Sync: Different paths within the topology can run on alternate schedules to prevent simultaneous WAN saturation across remote links.

Conflict Resolution

Because Windows 2000 DFS replication operates in a multi-master environment, concurrent edits to the same file across multiple replicas trigger automated conflict resolution. FRS resolves conflicts using a "last writer wins" rule based on the Coordinated Universal Time (UTC) timestamp of the modification. The losing version of the file is moved to a hidden Conflict.ntfrs folder located within the local staging path.