Establishing surveillance for XRF

Use the following five parameters to establish surveillance for XRF: SURV, LNK, LOG, RDS, and SWITCH.

1. Determine which surveillance mechanisms to use for the complex.

   Use the SURV parameter to establish the surveillance mechanisms. You can specify combinations of the three mechanisms: LNK, LOG, and RDS. Because of the importance of surveillance, however, you should use all three mechanisms.

   The alternate IMS uses these three surveillance mechanisms in different ways. Failure of the alternate IMS to receive signals on the ISC link or the RDS is possible cause for takeover; failure of the alternate IMS to receive new records on the log is not. Rather, IMS uses the LOG option to confirm or override a decision that the alternate IMS might make based on lack of signals from the RDS or the ISC link. For example, suppose you specify SURV=(LNK,LOG), and the alternate IMS stops receiving signals on the ISC link. The failure of signals over the ISC link indicates a takeover, but as long as the alternate IMS continues to receive log records satisfactorily, it does not request a takeover. In this case, IMS assumes that the ISC link itself is experiencing difficulty and is therefore not a reliable indicator.

   In an XRF complex that uses USERVAR, if you do not have an ISC link already in place, establish it through the 37x5 Communications Controller that controls your class-1 terminals.

   The surveillance mechanisms and the internal parameters can be changed dynamically by commands issued from the master terminal.

   Related reading: For more information on surveillance mechanisms, see XRF parameters in DFSHSBxx.

2. For each surveillance mechanism you choose, decide the interval value (how often the alternate IMS is to receive the signals).

   After you have established which surveillance mechanisms are to operate in the complex, you should decide how often the alternate IMS should receive signals through the surveillance.

   Code the LNK and RDS parameters to establish the intervals between signals for the LNK and the RDS for each of the surveillance mechanisms you choose.

   Two factors to consider in setting the timing for LNK and RDS are:

   • The speed of communication over the ISC link.
   • The performance overhead on the alternate z/OS® operating system. Set the timing interval higher if you need to reduce the work of the alternate z/OS operating system.

   Code the LOG parameter to establish how often the alternate IMS should check the log for new records.

   For more information about interval values, see the topic XRF parameters in DFSHSBxx.

   Related reading: IMS has specific rules and recommendations for coding the interval values for the two systems. See IMS Version 15 System Definition for these rules.

3. Decide whether the absence of any or all of these signals should be a takeover condition.

   Use the SWITCH parameter to establish absence of surveillance signals as criteria for takeover. Specify on the SWITCH parameter the surveillance mechanisms that are critical in alerting the alternate IMS of certain events. For example, an installation managing its local locks on the active IMS with IRLM, or one using data sharing, would specify the IRLM failure as a criterion for takeover. Installations with a large VTAM® network would probably specify a takeover to occur when a VTAM failure causes an IMS TPEND exit. An installation with a large non-VTAM network and a small VTAM network would probably not specify a VTAM failure as a criterion for takeover. For more information about the SWITCH parameter, see the topic XRF parameters in DFSHSBxx.

4. If the answer to 3 is yes, decide the timeout value (the length of time the alternate IMS waits for a surveillance signal before considering a takeover).

   After you have specified which surveillance signal absences are takeover conditions, you can decide how long the alternate IMS should wait for a signal before requesting a takeover. Specify timeout values on the LNK, LOG, and RDS parameters. The default intervals are 9 seconds for the LNK parameter and 3 seconds for LOG and RDS.

   If timeout values are too low, some z/OS recovery routines that are transparent in a non-XRF environment might cause takeovers. An example of such a routine is an alternate CPU recovery (ACR) routine. These unwanted takeovers are more likely to occur on a larger CPC or with a large IMS workload.