For the first scenario, a data seeder was utilized to create tasks within Alumio. We eliminated the influence of external factors by utilizing a data seeder instead of an API call, database query, GraphQL query, etc. With this approach, we were able to measure the overhead added by Alumio.

The following components were used while conducting the test: an incoming configuration, range Subscriber (to create 1000 entities per run), an entity transformer, a data filter, move using a pattern (data transformer), and a value setter.

For this scenario, a storage publisher was utilized to process tasks within Alumio. This time, we eliminate the influence of external factors by utilizing a storage publisher instead of an API call, database query, GraphQL query, etc. Since each task gets processed sequentially, there was no need for optimizations. However, it is possible to further improve the performance by executing bulk processing or upgrading the hosting environment.

‍

The following components were used - an outgoing configuration, storage publisher (to process 1000 tasks sequentially per run), an entity transformer, a data filter, move using a pattern (data transformer), and a value setter.

A synchronous integration approach creates and processes tasks at the same time. As data gets imported (tasks are created), it is immediately exported (tasks are processed). Similar to the asynchronous strategy, we used a data seeder to import the data and a storage publisher to export the data. This eliminates the influence of external factors and we measured the overhead added by Alumio.

‍

The following components were used - an incoming configuration, a range subscriber (to create 1000 entities per run), an outgoing configuration, a storage publisher (to process 1000 tasks sequentially per run), an entity transformer, a data filter, move using a pattern (data transformer), and a value setter.

‍

Asynchronous Approach

Average time taken for 1000 tasks: 42s 776ms

Median time taken for 1000 tasks: 44s 490ms

Average tasks per second: 23.378

Median tasks per second: 22.478

‍

Synchronous Approach

Average time taken for 1000 tasks: 4s 466ms

Median time taken for 1000 tasks: 4s 374 ms

Average tasks per second: 223.914

Median tasks per second: 228.623