Abstract: An independent distributed database system comprising a plurality of sites wherein all users at all sites work off-line with local data. All application transactions are against the local database only, and every site stores “all and only” the data it needs. On-line transactions occur only in the background, including a periodical “synch” between sites that transmits any changes to data of interest to that site. If the background operations are interrupted or the network is temporarily unavailable, the user does not see new changes made at other sites until the data link is available again, but is otherwise unaffected. It is a feature that no site acts as a “server” for any other site. Some sites may store more data or have more users than others, but all sites are logically peers.