How to use Google's large-scale graph processing architecture Pregel in ArangoDB

Interesting article. What kind of graph algorithms can be implemented in this Pregel framework and how compares the complexity of this approach to specialised algorithms not using the framework? Nay hints or references would be appreciated.