(This page uses style sheets.)
Any migration exercise should consist in general of:
A data gathering and project definition phase including:
A description of the set of relevant initial conditions consisting of, for instance:
systems architecture(s),
applications and their associated data,
protocols and standards used,
hardware,
the physical environment, such as network bandwidth, location,
the social requirements such as language(s) and staff skill sets;
A set of target conditions in the same detail,
A description of how to get from the existing to the planned conditions;
A justification for the migration including the cost associated with it,
One or more pilot phases which are designed to test the plan and the justification. Data from these pilots can then be fed back into the cost model used in the plan,
Roll out of the plan,
Monitoring of the actual experience against the plan.
The contents of item 1 above define something which in these guidelines is called a Scenario and the guidelines describe how to migrate to OSS in such circumstances.
However, for the guidelines to be readable and useful in practice there have to be some simplifying assumptions, otherwise the total number of possible combinations would become unworkable.
We have chosen one of the many different target environments (1B above) and have simplified the description of the initial environments (1A above). The target environment is discussed in Section 8.2. With this standard target environment assumed, a Scenario becomes defined by reference to the simplified initial environment and the migration path from it to the target.
New chapters can be added to this document as necessary in the future. Starting with Chapter 14, each chapter provides a reasonably detailed description of one Scenario, describing how to migrate to the target including discussion of partial migrations. Each chapter should be updated as experience is gained of real migrations.
In addition, there is an associated spreadsheet to help with item 2 above. In this spreadsheet the cost of each Scenario can be compared to the cost of the target together with the cost of migration.
The detail available in the survey of publicly available case studies was very limited. Only a small number of such studies were found (a list is given in Appendix A) which provided any detail at all other than a general press release. This means that most of the guidelines are based on the experience of netproject and its consultants and their discussions with people who have done a migration but not published their results.
The huge number of different combinations of initial and final conditions, together with the many different ways of getting from the one to the other, means that it is impossible for any set of guidelines to cover all the possibilities. The guidelines must therefore be considered as indicative of what can be done rather than prescriptive of what should be done. They should be used as a starting point in the process of migration. They should not be expected to provide an answer in all circumstances.
The migration is assumed to have a target which is a totally OSS environment where possible and sensible, however, there may be many reasons why proprietary systems may need to be kept or used. The possibility of a partial migration is also discussed.