Managing the number of server instances per server address space
A server address space contains one or multiple server instances which all select work requests from a work queue. Without using the new interface IWMSINF, the server address space must tell WLM during start up how many server instances will be started (PARALLEL_EU parameter on IWM4CON).
The IWMSINF interface allows the application to obtain the number
of server instances from WLM. Figure 1 shows
an example of how the server manager address space can use IWMSINF.
- To use the IWMSINF interface, the server manager address space
connects to WLM with IWM4CON specifying the
MANAGE_TASK=YES
option. All server address spaces that connect to the same application environment must specify the same setting of MANAGE_TASK. TheSERVER_MANAGER=YES
andSERVER_TYPE=QUEUE
options must also be defined. - After connecting, the server manager address space must invoke the IWMSINF service to obtain the number of server instances to be started initially.
- The server address space starts the server instances which select work from WLM by using IWM4SSL.
- The server manager address space again invokes IWMSINF to listen for recommendations from WLM for the number of tasks to start.
- IF WLM wants to terminate one or multiple server instances, it resumes IWM4SSL with a new return code telling the server instance to terminate.
- If WLM wants to terminate the server address space, it resumes IWM4SSL and IWMSINF with a return code that tells the server to terminate.
Note: WLM can only manage the number of server instances per server
address space if there is a linear relationship between the number
of instances in an address space and the virtual storage consumed
in the address space. This implies that all instances allocate about
the same amount of virtual storage.