There are also a few ways to spread make managed processes over multiple nodes, assuming they share storage, which could be useful if your transforms bottleneck

There are also a few ways to spread make managed processes over multiple nodes, assuming they share storage, which could be useful if your transforms bottleneck at the CPU rather than on storage or network IO.