On May 14, 2007, at 21:33, Pierre Demartines wrote:
Ideally, what I need is something like BigTable, i.e. a permanent
distributed storage for real-time data serving. It needs to be
reasonably fault-tolerant (say a server goes down for a while or a
disk
goes bad; the system needs to keep working in the meantime, and
when the
server comes back up, its own contribution to the distributed cache
needs to be repopulated transparently). New servers can be added, old
ones removed.
You're looking in the wrong area, then. Lucene's got Hadoop which
is providing some of the fundamentals, but it needs a bit of work.
Hadoop has a distributed filesystem and mapreduce, but from what I
can tell, they don't yet have a bigtable implementation.
I'm sure they'd appreciate having someone finish it. :)
--
Dustin Sallings
