Martin Fandel wrote:
> Which is the best solution? (Slony, DRDB, WAL or something else?)
Slony is a master/slave replication system that can improve overall read
performance. DRBD gives you a hot stand-by system. Both of these are
completely different goals, which can be implemented complementarily.
A WAL-based replication gives you about the same effect as a DRBD
replication, but with DRBD you pass on all the responsibility to the
operating system. One guy at LinuxTag told be he manages his DRBD
pairs using Linux virtual servers, which would mean that on some level
you don't even notice that you have two machines running instead of
one. If have yet to figure that out, but it sounds quite nice.
Btw., it was probably me who told you about DRBD, but I didn't make that
up myself. A lot of people are currently using it.
--
Peter Eisentraut
http://developer.postgresql.org/~petere/
