Search System Planning and Management Services
There are seven steps in implementing an enterprise search system. These seven “modules” are:
- Pre-project planning and set up
- Gathering basic information for requirements
- Spelling out the requirements
- Procuring the search system
- Deploying the search system
- Tuning the search system
- Operating the search system.
(See an overview of the steps in an enterprise search deployment.)
Each of these modules of work contains subtasks. A good example is tuning the search system once it is deployed. Tuning consists of verifying that security and content licensing functions are operating properly. Improving the performance of the search system almost always include improving the speed of returning results to users. Without proper planning, improving query response may require significant changes to other parts of the search system. One common “fix” is to reduce the frequency of index updates in order to free resources to process user queries. The result is that results may display faster but the information displayed may be stale.
There are not one or two trade offs in enterprise search systems. There are dozens, sometimes hundreds. Without appropriate planning, a decision made to address one aspect of the system may have significant and largely unexpected consequences in other parts of the system.
Arnold IT has numerous examples of individuals trivializing the planning activity for enterprise search systems. In the oft-quoted words of a prominent U.S. government official, these professionals “don’t know what they don’t know.” The problem of “unknown unknowns” with regard to enterprise search become evident only when costs sky rocket unexpectedly.
A good rule of thumb is to plan slowly and thoroughly, then execute quickly. Enterprise search systems are viewed as easy-to-buy, easy-to-operate systems. The reality is that search systems and their subsystems deal with the use of language. Nothing is more subtle or complex than this aspect of human behavior. Computers and software are better at repetitive tasks and counting than interpreting what information a person wants, requires, or needs.