To clarify. We have *no* XCF connection between primary and backup data centers. All DASD is mirrored continuously via XRC, but the DR LPARs are 'cold'. They get IPLed only on demand: for (frequent) testing and for (godforbid) actual failover.
When got into serious DR in the 90s, channel technology was ESCON, and network technology was ISV CNT. Parallel sysplex synchronization was governed by external timers (9037). When we started with parallel sysplex, loss of timer connection would kill the member that experienced it first. Then IBM introduced a change whereby the entire sysplex would go down on timer loss. This technology did not bode well for running a single sysplex over 100+ KM. Network connectivity was far too flaky to bet the farm on. Now we have FICON over DWDM. Way more reliable, but sysplex timing would still be an issue AFAIK. In our actual sysplexes (prod and DR), boxes are literally feet apart connected by physical cables du jour. I cannot recall a complete loss of XCF connectivity ever in this configuration. I'm still not clear on how a 'geographically dispersed sysplex' (original definition, not 'GDPS') would work. Critical data sets must be shared by all members. One of each set of mirrored pairs must be chosen as 'The Guy' that everyone uses. If The Guy suddenly loses connection to the other site--i.e. site disaster--how will the surviving member(s) at the other site continue running without interruption? If there is an interruption that requires some reconfig and IPL(s), then what's the point of running this way in first place? We commit to a four-hour recovery (including user validation) with data currency within seconds of the disaster. . . J.O.Skip Robinson Southern California Edison Company Electric Dragon Team Paddler SHARE MVS Program Co-Manager 323-715-0595 Mobile 626-543-6132 Office ⇐=== NEW robin...@sce.com -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of retired mainframer Sent: Thursday, January 11, 2018 9:20 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: (External):Re: SYSPLEX distance > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] > On Behalf Of Rob Schramm > Sent: Thursday, January 11, 2018 9:01 AM > To: IBM-MAIN@LISTSERV.UA.EDU > Subject: Re: SYSPLEX distance > > SFM and planning for what your surviving system should always be done. > And yes early on there was a failure of one of the two dark fiber > connections and the sysplex timers were not connected properly to > allow for a continued service. > > Planning planning planning. To which you should add testing testing testing. And once the developers of the plan have succeeded in making it work, it should be tested again with many of the least experienced people in the organization. Murphy will guarantee that they will be the only ones available when it really hits the fan. (It is amazing how differently a pro and a rookie read the same set of instructions.) ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
