ignite-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From vkulichenko <valentin.kuliche...@gmail.com>
Subject Re: Ignite terminology
Date Mon, 28 Mar 2016 21:37:02 GMT

Every time you call Ignition.start(..), a new node is started and a new
instance of Ignite is created. If you start several instances in the same
JVM, they are required to have unique names, which you then can use to
acquire concrete instance using Ignition.ignite(..) method. (Note that there
is no requirement for all nodes in one topology to have the same name.
Whether nodes discover each other or not is defined only by discovery SPI.)

There are two typical scenarios when this can be used:

1. You start two or more Ignite instances that connect to different
clusters. In this case each instance will give you access to different
cluster, so you can treat instance names as local references to these
2. You start the whole topology within one JVM. This is very useful for unit
testing and debugging. You can take a look at any of Ignite's internal
tests, we utilize this approach heavily.

In all other cases you should not worry about this setting at all and simply
use 'null' as the instance name (which is already the default value).

Makes sense?


View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Ignite-terminology-tp3687p3723.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

View raw message