System Load and Redundancy
Each location that is involved in your Watson™ Explorer Engine Platform can contain large amounts of data that may be heavily used at certain times. Due to the modular nature of Watson Explorer Engine, portions of the platform can be divided across multiple installations of Watson Explorer Engineto distribute the load. The querying and processing (front-end) can be separated from the indexing and retrieval (back-end), queries can collect results from multiple installations, and very large indices can be distributed across multiple installations. Additionally, Watson Explorer Engine supports multiple load-balanced front-ends and load-balanced index synchronized back-ends. Exactly how you should deploy this will depend on your user load and data source sizes. Figure 1 gives an example of a front-end/back-end system using mirrored front-ends and synchronized backends.