Decision Trees for Collection Partitioning

As discussed in the previous sections, how to divide data indexing across servers and across collections on a given server is a standard decision that you should make during the planning and design phases of your Watson™ Explorer Engine platform application. The partitioning scheme that you select can have an immediate impact on the speed with which your Watson Explorer Engine platform application can retrieve query results, and can have a long-term impact on the frequency with which your search collections can be updated (refreshed), the time required for rebuilding your search collections by reindexing, and so on.

Like most powerful, flexible technologies, the Watson Explorer Engine platform does not place any restrictions on how you organize your search collections. The next two sections provide decision trees that are designed to help you identify appropriate search collection creation and deployment scenarios, based on the organization of the data that you are indexing, your hardware, and other site-specific concerns.