Oracle Distributed Systems

Geographic Data Distribution

As hinted in the previous section, many organizations have a requirement for global data distribution. In other words, it may be necessary to maintain the same data in separate data centers on machines that are probably not even on the same network. For example, a web site that tracks user data about news preferences may deploy databases in California, Chicago, and New York so that it can direct user traffic to a site that is nearby. Or a large corporation may replicate data about its products and prices to numerous regional headquarters.

Oracle’s advanced replication is the clear choice for meeting these requirements. It can provide read/write access to data in multiple locations. In addition, if asynchronous propagation is being used, data is always fully accessible locally regardless of the availability of the remote sites. Sites also can be added or deleted as requirements change.

The most significant issues with using advanced replication for geographic data distribution are data latency and conflicts. Since synchronous propagation is not a viable option for machines that are in different locations, every site will be out of sync with its peers by at least one second (the shortest propagation frequency possible). If your application is one that allows the same tables to be updated frequently from multiple locations, this latency may be detrimental. Consider, for example, a user of a web site who changes her news preferences while she is connected ...