Another way to handle concurrent access is to assign each thread / process its own data structure (a table row or an element of a “global shared” list for example) when it runs, and to accumulate the data when needed.