Most of standard parallel join algorithms try to overcome data skews with a relatively static approach.
The way they distribute data (and then computation) over nodes depends
on a data re-distribution algorithm 
(hashing or range partitioning) that is determined before the actual join begins.
   On the contrary we choose to pre-scan data in order to choose an
   efficient join method for each given value of the join attribute. By the way, 
   this approach already proved to be efficient both theoretically and practically in our former papers.

   In this paper we introduce a new pipelined version of our frequency adaptive join algorithm.
   The use of pipelining decreases the number of disk accesses and the number of synchronization steps.
   We present a detailed version of the algorithm and a cost analysis
   based on the BSP model, showing that our pipelined algorithm
   achieves noticeable improvements compared to the one-step version.
   We thus show that the frequency adaptive approach remains efficient for pipelined multi-join queries.

Key words :   PDBMS (Parallel Database Management Systems), Parallel
joins, Multi-joins, Data skew, Join-product skew, Dynamic load balancing.