| [[ServicePool-ServicePool]] |
| === Service Pool |
| |
| *Available as of Camel 2.0* |
| |
| Camel supports pluggable pools for services. At this time of write we |
| have pools for: |
| |
| * Producer |
| |
| The default producer service pool is |
| `org.apache.camel.impl.DefaultProducerServicePool` and is used by |
| default in Camel for pooling Producer. |
| |
| The need for pooling Producer is only apparent in some |
| link:components.html[Components] in Camel to support thread safe |
| producers and support concurrency. And using pooling we can improve |
| performance as we do not have the overhead of creating, starting and |
| stopping the Producer at each invocation. So where is it needed then? |
| Well these components uses pooled producers: |
| |
| * link:ftp2-component.adoc[FTP] |
| * link:mina-component.adoc[Mina] |
| |
| [[ServicePool-Usage]] |
| ==== Usage |
| |
| By default Camel uses a shared |
| `org.apache.camel.impl.DefaultProducerServicePool` provided by |
| `CamelContext`. This default pool will pool up till 100 producers per |
| link:endpoint.html[Endpoint]. So if you have 6 endpoints there can be up |
| till 6 x 100 in total in the pool, where each distinct |
| link:endpoint.html[Endpoint] have a limit of up till 100 producers. |
| |
| [[ServicePool-Pluggableproducerpooling]] |
| ==== Pluggable producer pooling |
| |
| Camel supports using a 3rd part pool implementation. What is needed is |
| to implement an adapter by implementing the |
| `org.apache.camel.spi.ServicePool` interface and define the generic as |
| `Endpoint` as the key and `Producer` as the service. |
| |
| Then set your custom pooling on the `CamelContext` with the |
| `setProducerServicePool` method. |
| |
| [[ServicePool-Developingpooleableproducers]] |
| ==== Developing pooleable producers |
| |
| The producer service pool identify a producer as being pool capable if |
| the producer implements the marker interface |
| `org.apache.camel.ServicePoolAware`. + |
| As the producer will be pooled and thus long lived, your producer |
| should be able to automatic safely recover lost connection. Usually you |
| implement logic that re connects if needed. |
| |
| [[ServicePool-SeeAlso]] |
| ==== See Also |
| |
| * link:architecture.adoc[Architecture] |
| |
