The coordinator node routes the request in parallel to all of the replicas and waits for responses. One of those replicas might happen to be the coordinator itself.
Only replicas read/write data they are responsible for, not the coordinator (unless the coordinator is also a replica for that data). - Tyler On Thu, Jan 13, 2011 at 12:07 AM, raoyixuan (Shandy) <raoyix...@huawei.com>wrote: > I mean whether both the coordinate node and the replica node keep the > insert data. Or just the replica node keep the insert data. And the > coordinate node just route the insert data to the replica. Can you get me? > > -----Original Message----- > From: Jonathan Ellis [mailto:jbel...@gmail.com] > Sent: Thursday, January 13, 2011 1:56 PM > To: user > Subject: Re: about the insert data > > On Wed, Jan 12, 2011 at 5:46 PM, raoyixuan (Shandy) > <raoyix...@huawei.com> wrote: > > So you mean the coordinator node is just responsible for routing the > request. > > Right. Of course, if the coordinator node happens to also be a > replica, it can be a little more efficient by performing that > operation directly rather than going over a socket. > > > where the request will be Routed? whether the coordinator node route the > request to the first replica to insert the data? > > I don't understand the question. > > -- > Jonathan Ellis > Project Chair, Apache Cassandra > co-founder of Riptano, the source for professional Cassandra support > http://riptano.com >