Bit-Based Replication

The bit-based replication process operates as a connectionless system for synchronizing file and database information between a device and a maximum of two external data stores. Data that has changed in the object store of a device can then be synchronized with data on another system.

There are two main areas of replication. First, each object in the object store, which only includes files, databases, and database records, has 4 bits of information that can be use to determine if an object has changed.

The following list shows how the 4 bits are assigned:

  • One bit is the other bit, which can be used for any purpose by an application.
  • Two bits are dirty bits, which are used to identify an object that has changed since it was last synchronized with the desktop.
  • One bit is a tri-state bit, which handles the synchronization of the dirty object as a transaction that can be rolled back.

Note that with only two dirty bits, you can synchronize an object with a maximum of two different desktops using the same synchronization engine or two different synchronization engines with each engine only synchronizing to one desktop. The synchronization bits should only be manipulated by the application that created the object. If an application manipulates synchronization bits of an object that they did not create, there is a potential for data loss.

The second area of synchronization is the notification of dynamic changes in the object store. This enables a synchronization engine to synchronize data in real time. The main issue is that only one synchronization engine can receive notification at one time.

Last updated on Wednesday, April 13, 2005

© 2005 Microsoft Corporation. All rights reserved.