Agreed. The reason the arangodb team give is the different concurrency models between arangod and node, but I don't think the problem is insurmountable. From the programmers perspective it would be no different from using node's cluster module.