Windows 2000 Active Directory Replication Compression

In Microsoft Windows 2000 Active Directory, replication mechanisms balance network bandwidth consumption against domain controller processing overhead. To manage this trade-off, the operating system distinctly separates intra-site replication (within a single local area network) from inter-site replication (across wide area networks), selectively applying data compression only where bandwidth conservation is critical.

Intra-Site Replication: Uncompressed for Speed

Intra-site replication is engineered under the assumption that domain controllers within the same Active Directory site are connected via fast, reliable, and cost-effective Local Area Network (LAN) connections (typically 10 Mbps or higher).

  • No Compression: Windows 2000 does not compress directory updates during intra-site replication.
  • CPU Preservation: Because local network bandwidth is abundant, consuming server CPU cycles to compress and decompress directory objects would introduce unnecessary latency. Leaving the data uncompressed minimizes CPU utilization and ensures rapid directory convergence.
  • Transport Protocol: It exclusively uses Remote Procedure Call (RPC) over IP.
  • Notification Model: When a change occurs, the source domain controller waits for a brief interval (default of 300 seconds, with 30-second delays between subsequent partners) and notifies its replication partners, which then request the uncompressed updates.

Inter-Site Replication: Compressed for Bandwidth Conservation

Inter-site replication operates across Active Directory sites connected by Wide Area Network (WAN) links, which are typically slower, prone to latency, and more expensive.

  • Mandatory Compression: Windows 2000 automatically compresses all directory updates transmitted between sites. The compression reduces the payload size by approximately 85% to 90% (transmitting roughly 10% to 15% of the original uncompressed volume).
  • Processing Overhead: To achieve this bandwidth reduction, Windows 2000 utilizes the Lempel-Ziv-based MS-ZIP compression algorithm. The bridgehead server at the sending site compresses the update packets, while the bridgehead server at the receiving site decompresses them before committing them to the local ntds.dit database. This trades higher CPU utilization on bridgehead servers for significantly lower WAN link usage.
  • Transport Options: Inter-site replication can occur over RPC over IP or Simple Mail Transfer Protocol (SMTP). Regardless of the transport protocol chosen, the underlying directory data payload is compressed prior to transmission.
  • Schedule and Polling: Unlike intra-site replication, inter-site replication does not use immediate change notifications by default. It relies on a defined schedule to consolidate updates into larger batches, optimizing the efficiency of the compression algorithm.