With High Concurrency we don't necessary mean several tens of users performing data calculation upon the same measures (info-cubes) but also the capability to simultaneously perform operations, such as data-readers that are traditionally queued. Here above, you can find some highlights about those changes of behavior you have and that may steer your single procedure or entire data workflow design. All what is mentioned is of course applicable when two or more users perform operations on the same data-set; the behavior of a procedure when performed by a single user is unaltered.
Data-entry and Data-reading are performed simultaneously if the info-cube is dense or if the data-reader updates the info-cube dimensions (entities) only
Data-flow and Data-reading are performed simultaneously if the info-cube is dense or if the data-reader updates the info-cube dimensions (entities) only
Data-flow and Data-entry are performed simultaneously
Data-readers are performed simultaneously even if the update the same database object
Data-flows are queued is the target info-cubes are sparse and they share the same sparse
Info-cube clear is always queued unless you run it using the current selection
A Procedure Sample
If two or more users are simultaneously triggering a Procedure, performing data-flows where the source info-cube is previously loaded with a data-reader, it can happen that the data-flows work on a data-set that is still incomplete, since the data-reader is still running. I.e. the data-reader and the data-flow are executed simultaneously.