Fig. 1

The Azure pipeline for running the benchmark. The GeometryTasks queue is loaded with one message per geometry pair. The GeometryHandler processes each message, by fetching the two geometries in the pair and storing them in a blob store. It then puts four messages on the BenchmarkTasks queue, one for each algorithm under test. This message has a reference to the blob id of the geometry pair. The benchmark function then loads the geometry pair and executes the three operations (create, apply and, update) using the specified algorithm, while logging time usage. The results are then stored in a table storage for later retrieval and analysis