General Computation

It is possible to use Speedment as a General Computation engine whereby results are computed between several Speedment instances. Each successive step will read directly from memory from one or more previous stages instead of reading from a regular database.

In the table below, the steps 0 - N are illustrated where the initial step 0 is populated from a database:

Step # Data Source Action
0 Database Step 0 (the initial step) reads from a general data source as per default a database
1 Step 0 Step 1 reads input data from Step 0
2 Step {0-1} Step 2 reads input data from a set of previous steps {0, 1}
N Step {0..N-1} Step N reads input data from a set of previous steps {0, 1, …, N-1}

The General Computation engine features will be improved over the course of the coming releases.


By replacing the SqlStreamSupplierComponent in a step, that step can be made to read from previous stages instead of reading from a database. The method of using Custom Stream Suppliers is generally described here.

The following example shows how a simplistic two-step computation engine can be implemented for the Sakila Database. The only modification of inter-step-data that is carried out is that the rental rate for Film entities is multiplied by 10 compared with data from the previous step.

public class SimpleTwoStepExample {

    public static void main(String[] args) {

        // The First Step Reads From the Database
        final SakilaApplication step0 = new SakilaApplicationBuilder()


        final FilmManager filmsStep0 = step0.getOrThrow(FilmManager.class);
        // Print out the first 4 Films, Data was read from the database;

        // The Second Step Reads From the firstStepApp in-JVM-memory data
        final SakilaApplication step1 = new SakilaApplicationBuilder()

        step1.get(StepStreamSupplierComponent.class).ifPresent(p -> p.setPreviousStage(step0));

        final FilmManager filmsStep1 = step1.getOrThrow(FilmManager.class);
        // Print out the first 4 Films, Data was read from the first step;


    private static class StepStreamSupplierComponent implements SqlStreamSupplierComponent {

        private SakilaApplication previousStageApp;

        public <ENTITY> Stream<ENTITY> stream(TableIdentifier<ENTITY> tableIdentifier, ParallelStrategy strategy) {
            final StreamSupplierComponent ssc = previousStageApp.getOrThrow(StreamSupplierComponent.class);
            if (Film.FILM_ID.identifier().asTableIdentifier().equals(tableIdentifier)) {
                return, strategy)
                    .map(f -> f.setRentalRate(f.getRentalRate().multiply(BigDecimal.TEN)))
                    .map(f -> (ENTITY)f);
            return, strategy);

        public void setPreviousStage(SakilaApplication previousStageApp) {
            this.previousStageApp = requireNonNull(previousStageApp);


It should be noted that in a more realistic example than above, the data model may be different for each step and the stream content may also be modified such that the content of several tables are modified. In such cases, the same principle as shown above can be applied but the implementation of the custom stream supplier’s stream() method must be tailored to match the input requirements of the new stage.

Questions and Discussion

If you have any question, don’t hesitate to reach out to the Speedment developers on Gitter.