This paper is just a thinly-veiled sales pitch for Vertica. All of the C-Store players are on it. Stonebraker has officially jumped the shark.
This is academically interesting but probably not of much practical use since parallel relational databases are not free (and may never be).
It probably won't take too long for someone to implement an open source version of Greenplum using Postgres. And there are a lot of people who are willing to pay for high-performance analysis of large data sets (as the success of Teradata, Netezza, et al. confirms).