The visible part of Harvest-plan is the web application that harvest planners (and their harvest partners) interact with using via desktops, tablets or smartphones. This is used to gather vineyard data and provide access to the maps and analytics based on that data.

Supporting the web tier is an integration tier, the role of which is to manage and orchestrate data flows to and from all planners, data sources and specialist services. This is the busiest part of the system and the place where we install the smart components that need to be immersed in all the data flows they learn from. Specifically, this is where the PoolEngine and PredictionEngine reside.

Underpinning it all is the data tier, which we call the “pool” (the “Thoughtpool” we named ourselves after) because it contains the sum of all data. It comprises databases and files, and the main job of the PoolEngine is to ensure all this data exists as a coherent whole (so it can be analysed scientifically) while keeping all data slices neatly separated.

The next couple of posts will deal with our prediction strategies.