FEWS Set-up - Local vs Client-Server
There are two ways to set-up a FEWS system, as a local system or a client-server. A local system is also known as Stand Alone (SA), a Client-Server set-up can also be referred to as Live System. Please be aware that within the scope of RISC-KIT only the set-up of Stand Alone systems are foreseen.
The configuration of a Local and a Client-Server system are the same. The user (or client) uses either a Stand Alone application or an Operator Client. The look and feel and functionality (import, pre- and post-processing, running of simulations, displaying results) of these applications however are highly similar. The difference is in the back-end, or on the server. In a Stand Alone environment, there is no server. In a Client-Server environment, the server side holds the Central Database, the forecasting shell(s) and the Master Controller who controls it all (see figure).
Local
In a local system, all data is stored in a LocalDataStore. This has as a consequence that simulations can be run one at the time, on 1 location only. This can be your local PC or a cluster, but not several locations at once. Workflows are executed serially either manually by the user or via something like Windows Task Scheduler. The set-up of a forecasting system always starts in a Stand-Alone environment.
Figure: Stand-Alone overview
Client-Server
In a Client-Server system, all data is stored in a Central DataBase. This has as a consequence that simulations can be run simultaneously on several machines, or so called forecasting nodes. Also, the model results can be accessed by multiple users at once. Each forecasting node has access to read and write all data available in the Central DataBase, whether it was generated by "himself" or one of the other forecasting shells. All tasks are scheduled to run automatically and managed by the Master Controller component.
Figure: Client-Server overview
