Streaming Support for Data Intensive Cloud-Based Sequence Analysis
Figure 4
Streaming/pipelining in workflow systems: (a) no pipelining, where A starts computation only when all its input items are available, and so does B which processes the output items of A. (b) Pipelining, where A starts computation when any data item is available, and so does B. Note that in this mode A and B run concurrently on different data items.
