WLM routing services give recommendations which cause routers to send too much work to certain systems. These situations are found on LPAR systems, very often with a load of around 95% and more than one logical processor assigned to the LPAR. The summed free capacity values across all LPARS contains, in this case, a value which is much higher than the available CEC capacity. This incorrect calculation of free capacity results in WLM making bad recommendations.

To deal with this, the Measured Waittime is put in perspective with the real available system and partition capacity. Thus, over-estimation isreduced. This new option on routing services allows a router to receive weights which take into consideration the goal achievement of the work.

Additionally, TCP/IP only registers the stack and doesn’t know how the work is being processed. A new routing service for TCP/IP has been introduced which allows the provision of routing recommendations for TCP/IP backend applications. However, abnormal conditions of the work receiver or consumer are not factored into the routing recommendations.

The queue time of the work is accounted for DDF regions, and its proportion to the execution time is factored into the routing recommendations.