By Frank Cohen

Many smart and well-meaning people in the information technology industry are searching for ways to understand the impact service oriented architecture (SOA) and master data management (MDM) will have on enterprise IT strategies. SOA and MDM are similar in one way: A stan-dards body manages neither. Consequently opinions about these efforts are widespread and var-ied. One thing that unites both is a new willingness among CIOs and data architects to build sys-tems that assert only a portion of the truth.

Base Computing is a methodology and set of patterns to govern development and operation of a set of business processes in a data center. Base Computing is the opposite of Enterprise Applica-tion Integration (EAI) in that Base Computing has no system of record, no central registry of software and services, and no central governance. To explain this in greater depth consider these axioms of Base Computing:

A base is integral to itself, so anything the base says is true is true from the base's perspective. There may be several bases with different values of the same data. It is up to the consumer of the base's service to determine which base to believe holds the truth. For instance, a package deliv-ery service depends on an airline cargo company and a trucking company to deliver a package to a consumer's premises. The air cargo company and trucking company operate bases that provide standards based interfaces to answer consumer requests such as "where is the package now?"

The air cargo company and trucking company bases interoperate to share package location in-formation. If the consumer requests package location from the air cargo company while the package is still on the truck then the air cargo company base asserts the most recent status of the package as received from the trucking company base. The consumer may also request package location from the trucking company. Base Computing does not expect the package location re-sponses to always be the same. In Base Computing environments it is up to the consumer to solve the differences.

Bases build trust relationships with consumers by working diligently to deliver the best possible data to the consumer. A base holds data in its own internal persistence store. The base uses pub-lish/subscribe, replication, and XA transactional protocols to keep the data it its store current. The consumer depends on the base to assert the validity of the data.

The consumer to a base may be a service or a human being operating a consumer application like a Web browser. The base provides a repository of the functions, interfaces, and workflows avail-able within the base. The consumer binds to the base using one of the protocols supported by the base. Some bases provide protocol and data format transformation between a consumer and an-other base.

Bases rarely provide 100% of a consumer's needed functions. Instead, Base Computing focuses on providing just enough functions and works hard to make certain those functions are high qual-ity from the consumer's perspective.