Dataset<Row> fullData = sparkSession.read().json("some_path");
ExecutorService executor = Executors.newFixedThreadPool(10);
List<Runnable> tasks = Lists.newArrayList();
for (int i = 1; i <= 50; i++) {
    final int x = i;
    tasks.add(() -> {
        Dataset<Row> subset_1 = fullData.filter(length(col("name")).equalTo(x));
        Dataset<Row> subset_2 = fullData.filter(length(col("name")).equalTo(x));
        Dataset<Row> result = subset_1.join(subset_2, ...);
        log.info("Res size is " + result.count()); // force Spark do the join operation
    });
}
CompletableFuture<?>[] futures = tasks.stream()
    .map(task -> CompletableFuture.runAsync(task, executor))
    .toArray(CompletableFuture[]::new);
CompletableFuture.allOf(futures).join();
executor.shutdown();