The Thunderbolt Paradigm

The Thunderbolt Paradigm The Thunderbolt Paradigm

The Thunderbolt view of data processing is that all problems can be reduced to three fundamental steps:

                1. Isolate: determine what data you are going to work with
                2. Arrange: how you want the data arranged
                3. Operate: what calculation, operation, and/or display is to occur.

Thunderbolt views this as a three-step process, a natural flow for problem solving. It treats them discretely and offers several commands at each step. Performance optimization is natural because each step sets up the next.

Other approaches attempt to lump what is essentially a three-step process into one convolution. With these approaches, simple problems become hard and the type of performance optimization we are talking about is simply non-existent.

The physical construct that makes the three-step process a reality is the collection. So the style of the Thunderbolt Query Language becomes something like:

                Find A ... (where something is true)
                Sort A by ... (a list of fieldnames)
                Report A ... (according to some report name)

or, in multi-table situations, this becomes

                Find A ...
                Match A ...
                Relate A ...
                Report A ...

In the above examples, A is the collection. The idea is to first determine the subset of interest and then operate on it. To paraphrase the FIND command, it is like saying "Find the subset that I am arbitrarily calling A for which the following conditions are true." Subsequent commands, by referring to A, are desensitized from the size of the database because all they see is the subset of interest.

We are, of course, only touching the tip of the iceberg here. As you begin to see very complex data processing problems break into this flow, you will come to appreciate it.