Load Balancing
A load balance system consists of at least one Lobster_data node controller and one Lobster_data node. The number of Lobster_data nodes is unlimited. It makes sense to install the Lobster_data node controller in a High Availability Environment, but it is not necessary. Even if it is not necessary to have a homogeneous operating system environment, it is recommended to run the Lobster_data nodes on the same operating system the Lobster_data node controller is running on. It is also recommended to use Unix/Linux operating systems, although it is not required.
Important note: Make sure that the message port (default is 8020) is open on all participating systems, i.e. adjust your firewalls and/or virus scanners. If the failover concept is activated, usually another port has to be open (see there).
General Remarks
The Lobster_data node controller receives the data of all input agents and delegates the resulting jobs to the active Lobster_data nodes. The only exceptions are all HTTP based input agents (e.g. web service) since they are built on a request-response model.
The GUI of Lobster_data can only log into the node controller. The entire activities of all the Lobster_data installations (node controller and nodes) are collectively displayed in the Control Center. All the active Lobster_data nodes are notified and synchronised with modifications on profiles (including the creation of profiles). Since all the instances use the same database, subsequently started nodes will load the latest profiles from the database and therefore be automatically synchronised.
Unless there is a different configuration (e.g. routes), the execution of jobs resulting from a profile chain will remain with the active Lobster_data node that has been assigned with the initial job. This reduces the network load and allows for a faster conversion of the data since the overhead of the distribution logic is avoided. Note: Alternatively, the message in the response route of a chained profile can be set to context DataWizard and queue LoadBalanceTask. This will force the message to be sent to the node controller, from where it will be distributed. Make sure not to set the IP 0.0.0.0 in the configuration file message.xml (controller and node)!