Data doesn't exist only at a point in time. Particularly for long-running production workflows, you are likely to acquire a significant quantity of data in a Hadoop cluster. Requirements rarely stay static for long, so alongside new logic you might also see the format of that data change or require multiple data sources to be used to provide the dataset processed in your application. We use the term data lifecycle management to describe an approach to handling the collection, storage, and transformation of data that ensures that data is where it needs to be, in the format it needs to be in, in a way that allows data and system evolution over time.