z/OS Communications Server: SNA Network Implementation Guide
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Performance consideration

z/OS Communications Server: SNA Network Implementation Guide
SC27-3672-01

By default, the UNRCHTIM start option has a value of 0 for all virtual routing nodes; therefore, EE connection network reachability awareness is disabled. While EE connection network reachability awareness can be enabled for all virtual routing nodes defined at this node (by this VTAM®) by setting the UNRCHTIM start option to a value in the range 10 - 65535, the preferred mechanism is to set UNRCHTIM at the GROUP level of the EE XCA major node. This setting provides maximum flexibility in choosing the unreachable time value for each VRN, and in selectively enabling and disabling the function for each VRN.

When EE connection network reachability awareness is enabled for a VRN, APPN Topology and Routing Services begins maintaining unreachability records if failures occur while using that VRN. The presence of unreachability records in the topology database increases the number of CPU cycles that are consumed during APPN route calculation. This CPU cycle increase is magnified as the number of VRNs that contain unreachability records increases and as the average number of unreachability records increases for each of those VRNs.

To mitigate the CPU cost of enabling EE connection network reachability awareness, the following considerations apply:
  • Typical EE network designs require that only few connection networks be accessible from a single APPN subnet. For network designs that employ more than four EE connection networks that are accessible from a subnet, EE connection network reachability awareness should be enabled for no more than four of those EE VRNs at any given time.
  • While the mechanism employed by EE connection network reachability awareness provides the tracking of connection failures on a per-partner basis, failures to many partners typically indicate a pervasive problem with the underlying transport. You can set a limit on the number of unreachable partner records (partner_limit) associated with a VRN on the UNRCHTIM start option. A network node maintains all unreachability records that are reported from its served end nodes and from all other network nodes in the same network. When the count exceeds this limit of unreachable partner records for a VRN, the VRN is considered unusable for any access until enough existing unreachable partner records expire or are manually cleared with the MODIFY TOPO,FUNCTION=CLRUNRCH command. The count of unreachable partner records must drop below 80% of the unreachable partner limit specified or defaulted on the UNRCHTIM start option for the VRN to again be considered for route selection.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014