For customer deployments we know that a high CPF workload will require more processing power and I/O than an average system. It would be beneficial to develop the following:-
BPMN2 processes and test flows to move those processes forward (must be parallel - unlike shtests scripts)
Batch test runner that increases load from 20 parallel processes to 40, 60 etc. with 15 minute intervals
Mathematical model and QConsole search queries that take the result processes' metrics data to calculate Workflow Tasks Per Second (WTPS) average completion times both for a) each type of step and b) average overall for entire set of tests
Modified Sizing Spreadsheet to include the 'workload' calculations and IO being done entirely on MarkLogic Workflow, and not anything else in the database (and thus can be combined with existing non-workflow sizing numbers).
NB Tests and models must therefore:-
Be of 'average complexity' - approx 20 steps per process model
Include all step types in a believable scenario (not just a lot of send email tasks, for example)
For customer deployments we know that a high CPF workload will require more processing power and I/O than an average system. It would be beneficial to develop the following:-
NB Tests and models must therefore:-