Parallel adoption

Parallel adoption is a method for transferring between a previous (IT) system to a target (IT) system in an organization. In order to reduce risk, the old and new system run simultaneously for some period of time after which, if the criteria for the new system are met, the old system is disabled. The process requires careful planning and control and a significant investment in labor hours.

This entry focuses on the generic process of parallel adoption; (real-world) examples are used for a more meaningful interpretation of the process if necessary. Moreover a process-data model is used for visualizing the process which is intended to provide a complete overview of all the steps involved in the parallel adoption, but emphasis will be laid on the unique characteristics of parallel adoption. Some common characteristics, especially defining an implementation strategy, that go for all four generic kinds of adoption are described in Adoption (software implementation).

Besides parallel adoption, three other generic kinds of adoption can be identified. The choice for a specific adoption method depends on the organizational characteristics; more insight on this topic will be provided below. The three other adoption methods are: Product Software Adoption: Big Bang Adoption (Also known as Direct Conversion, slam dunk, or cold-turkey strategy), Phased adoption and Pilot adoption.

There are several instances when parallel conversion can not be considered a viable conversion strategy. First consider if the new system contains significant schema changes. Data elements required by one system that are not being populated by the other can lead to at best data inaccuracies and at worst data corruption. Another concern is if the system relies on consumer off the shelf technology (COTS). If a COTS vendor's documentation states that more than one application can not share the same database, then parallel conversion is not an option. An example would be Oracle's Siebel products. Other COTS products may also place restrictions when patches or major upgrades require unique license keys. Once applied they may make database changes that might cause the application to falsely detect a parallel system running against the same database as an attempt at getting around licensing controls and thereby disable the system.

There seem to be little conventions regarding the process of parallel adoption. Several sources (e.g.: Turban, 2002, Eason, 1988, Rooijmans, 2003, Brown, 1999), do not use a single process-description name. The term parallel adoption is denoted in these sources, although consistent per source as: parallel conversion, parallel running, shadow-running, parallel cutover and parallel implementation. This appears to be the case because a generic description of the process does not need a distinct classification. There are a quite some standard implementation methods, where different adoption techniques are described but often in a practical context; real-world case scenario or a more comprehensive set of implementation techniques like Regatta: adoption method, SIM and PRINCE2. In general, parallel adoption can best be seen as a Systems Engineering method of implementation of a new system.

...
Wikipedia