On Dec 28, 2008, at 10:07 AM, Paul Davis wrote:
On Sun, Dec 28, 2008 at 9:54 AM, Geir Magnusson Jr. <[email protected]>
wrote:
On Dec 28, 2008, at 9:36 AM, Paul Davis wrote:
[snip]
I read the view docs (and have other questions there, like if the
M/R is
distributed across a cluster - I've used M/R w/ Hadoop, so I come
w/ a
set
of assumptions...) and I saw that it doesn't *appear* that the
key or id
is
injected in the view doc, which of course brings up an obvious
question
:)
The CouchDB implementation of Map/Reduce (specifically reduce) is a
bit different the the canonical implementation. Mostly this is a
result of being stored in a btree and Damien's cleverness.
Does the work get distributed across nodes, or does each node do
the same
thing?
Oh sorry, I totally meant to answer that question in my earlier reply.
No. Not yet. IIRC, the current rough sketch of a design is that
CouchDB will be able to do automatic sharding of a single database,
and each node would be responsible for building a view of its
particular shard. Then at query time you just have a big merge sort
type of operation.
Fantastic. What's the ETA?
geir
