IceStorm is a relatively lightweight service in that it requires very little configuration and is implemented as an IceBox service. The configuration properties supported by IceStorm are described in IceStorm Properties; some of them control diagnostic output and are not discussed here.
On this page:
IceStorm Property Prefix
As you will see in IceStorm Properties, IceStorm uses its IceBox service name as the prefix for all of its properties. For example, the property
DemoIceStorm.TopicManager.Endpoints when IceStorm is configured as the IceBox service
IceStorm Server Configuration
The first step is configuring IceBox to run the IceStorm service:
In this example, the IceStorm service itself is configured by the properties in the
config.service file, which might look as follows for a non-replicated service:
IceStorm uses LMDB to manage the service's persistent state, therefore the first property specifies the path name of the LMDB database environment directory for the service. Here the directory
db is used, which must already exist in the current working directory. This property can be omitted when the service is running in transient mode.
The final two properties specify the endpoints used by the IceStorm object adapters; notice that their property names begin with
DemoIceStorm, matching the service name. The
TopicManager property specifies the endpoints on which the
Topic objects reside; these endpoints must use a connection-oriented protocol such as TCP or SSL. The
Publish property specifies the endpoints used by topic publisher objects; using datagram endpoints in this property is possible but carries additional risk.
IceStorm's default thread pool configuration is sufficient when the service is running on a single CPU machine. On a host with multiple CPUs, you may be able to improve IceStorm's performance by increasing the size of its client-side thread pool using the
Ice.ThreadPool.Client.* properties, but the optimal number of threads can only be determined with careful benchmarking.
Deploying IceStorm Replicas
There are two ways of deploying IceStorm in its highly available (replicated) mode. In both cases, adding another replica requires that all active replicas be stopped while their configurations are updated; it is not possible to add a replica while replication is running.
To remove a replica, stop all replicas and alter the configuration as necessary. You must be careful not to remove a replica if it has the latest database state. This situation will never occur during normal operation since the database state of all replicas is identical. However, in the event of a crash it is possible for a coordinator to have later database state than all replicas. The safest approach is to verify that all replicas are active prior to stopping them. You can do this using the
icestormadmin utility by checking that all replicas are in the
IceGrid is a convenient way of deploying IceStorm replicas. The term replica is also used in the context of IceGrid, specifically when referring to groups of object adapters that participate in replication. It is important to be aware of the distinction between IceStorm replication and object adapter replication; IceStorm replication uses object adapter replication when deployed with IceGrid, but IceStorm does not require object adapter replication as you will see below.
An IceGrid deployment typically uses two adapter replica groups: one for the publisher proxies, and another for the topics, as shown below:
The object adapters are then configured to use these replica groups:
An application may not want publisher proxies to contain multiple endpoints. In this case you should remove
PublishReplicaGroup from the above deployment.
The next step is defining the endpoints for the adapter
Node, which is used internally for communication with other IceStorm replicas and is not part of an adapter replica group:
Finally, you must define the node ID for each IceStorm replica using the
NodeId property. The node ID must be a non-negative integer:
You can find a complete C++ example of an IceGrid deployment in the directory cpp
/IceStorm/replicated in the ice-demos git repository.
You can also deploy IceStorm replicas without IceGrid, although it requires more manual configuration; an IceGrid deployment is simpler to maintain.
The first step is defining the set of node proxies using properties of the form
Nodes.id. These proxies allow replicas to contact each other; their object identities are composed using
For example, assuming we are using the IceBox service name
IceStorm and have three replicas with the identifiers 0, 1, 2 and an instance name of
DemoIceStorm, we can configure the proxies as shown below:
These properties must be defined in each replica. Additionally, each replica must define its node ID, as well as the node's endpoints. For example, we can configure node 0 as follows:
The endpoints for each replica and ID must match the proxies configured in the
Two additional properties allow you to configure replicated endpoints:
Defines the endpoints contained in proxies returned by the topic manager.
Defines the endpoints contained in the publisher proxy returned by the topic.
For example, suppose we configure three replicas:
Each replica should also define these properties:
An application may not want publisher proxies to contain multiple endpoints. In this case you should remove the definition of the
ReplicatedPublishEndpoints property from the above deployment.
You can find a complete C++ example of a manual deployment in the directory cpp
in the ice-demos git repository.
IceStorm Client Configuration
Clients of the service can define a proxy for the
TopicManager object as follows:
The name of the property is not relevant, but the endpoint must match that of the
service.TopicManager.Endpoints property, and the object identity must use the IceStorm instance name as the category and
TopicManager as the name.
IceStorm Object Identities
IceStorm hosts a well-known object that implements the
IceStorm::TopicManager interface. The default identity of this object is
IceStorm/TopicManager, as seen in the stringified proxy example above. If an application requires the use of multiple IceStorm services, it's a good idea to assign unique identities to their well-known objects by configuring the services with different values for the
service.InstanceName property, as shown in the following example:
This property changes the category of the object's identity, which becomes
Measurement/TopicManager. The client's configuration must also be changed to reflect the new identity:
IceStorm also hosts an object with the identity
IceStorm/Finder, as described in the next section. This identity is not affected by the service name or by changes to
Using the IceStorm
IceStorm supports the
An object supporting this interface is available with the identity
IceStorm/Finder on the service's topic manager endpoint. By knowing the host and port of this endpoint, a client can discover the topic manager's proxy at run time with a call to