It’s very common, however, that data will reference the data to be deleted, or be derived from it in a way that will become invalid when it’s deleted. Parallel

It’s very common, however, that data will reference the data to be deleted, or be derived from it in a way that will become invalid when it’s deleted. Parallel processing becomes less straightforward because you have to make sure that all parallel processes see a consistent state of the deletion. Depending on the nature of the data to be deleted, you may actually have potential mutation, if there are keys that are sensitive data.