cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Greene <>
Subject Re: one server or more servers?
Date Wed, 15 Jul 2009 22:40:27 GMT
You can use the nodeprobe utility in bin/ to contact each node and
make sure they see the same information.  Run it with no arguments to
see the commands you can pass it.

There is also an open issue at for making this a
little more automatic (instead of having to run nodeprobe on each node
and check the results by hand, you can just pass in all the servers
that you think should be able to see each other) but there's no code
for this yet.


On Wed, Jul 15, 2009 at 5:33 PM, <> wrote:
> Mark and Jonathan
> I have started cassandra on 4 servers with all 4 of them as seeds
> how do i know all 4 are now part of the datastore
> are there ways to test this
> thanks a lot
> On Tue, Jul 14, 2009 at 8:30 AM, Mark Robson <> wrote:
>> 2009/7/14 <>
>>> 1. If you only have 3 production servers, Cassandra may not do much for
>>> you. You will probably only care if you have lots more servers. 3 servers is
>>> a reasonable minimum for a test / dev environment
>>> At How many servers does cassandra start really performing?
>>> or how many servers is an ideal setup say for a game 10?
>> It can use any number, three is probably a reasonable minimum, I don't
>> know what the maximum is.
>> If you have a small number of servers, there is probably no reason to use
>> Cassandra; you can more usefully use a conventional database with
>> replication or sharding.
>> The reason you might want to use Cassandra is to scale writes a lot, to
>> levels you couldn't reach with more "conventional" databases.
>> In all likelihood, an application would use Cassandra to store high-volume
>> high-write data alongside a more conventional database to store smaller,
>> less frequently changing stuff.
>> It appears to me that Cassandra 0.3 is only really useful if running on a
>> homogenous cluster of dedicated servers with reasonable amounts of memory /
>> storage (But not necessarily high spec servers of CPU, IO speed or internal
>> redundancy).
>> Once the load balancing is implemented in 0.5 (according to current plans)
>> it will become more useful for heterogenous environments.
>> Mark
> --
> Bidegg worlds best auction site

View raw message