I agree with Radoslaw that an elaborate CTC naming convention might be overkill for many shops. OTOH when we introduced sysplex in the mid-90s, we went from a handful of single-purpose LPARs across four CECs to multisystem plexes. We started with no CTCs to speak of, but XCF wants CTCs as backup for CF structures. Hence the number of CTCs ballooned in a short period. Our IBM Top Gun CE suggested a naming scheme because there were circulating stories of confusion and mishandling of CTCs, especially in error situations where an operator calls a sysprog at oh-dark-thirty to report a problem. Imposing some sanity on the chaos seemed like a good idea.
A few other comments. -- We run 14 LPARs in 'production', that is day-in day-out for public use. We also run an additional 7 LPARs during non-disruptive DR testing for a total of 21 LPARs spread across (now) three CECs in two data centers. Every LPAR has CTC connections to every other LPAR on every CEC in both data centers. -- We have never needed to configure more than 8 LPARs on a single CEC. Hence our naming scheme provides for both a primary and a backup range of addresses for each LPAR connection. The primary range goes through one FICON director, the secondary range goes through a different director. This doubles the number of connections while providing maximum redundancy. -- Using 'three digit addresses' really means using four digit addresses all beginning with '0'. I believe that's what we started with before sysplexing. That may well be adequate for many shops. Some naming scheme can be helpful for more complex configurations. . . 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 [email protected] -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of R.S. Sent: Thursday, October 18, 2018 9:19 AM To: [email protected] Subject: (External):Re: CTC conventions IMHO as every convention, this one is limited. You cannot put any number of CPCs or LPARs in it. And you don't need it. From the other hand any fixed-lingth field means some lost, overhead, i.e. One hex for CPC? That's to much for two CPCs in a shop and can be completely omitted for single CPC. The same for LPARs - do you really need more than 16 in CTC? Much more? How many bits? Of course it's also possible to define CTC with no reasonable convention at all. The only things which are really checked are CUADD number and UA for device. However all the devices can be numbered consecutively, as well as CU numbers. It would be nightmare to manage ...or not when using special table (and no changes). The advantage is absolutely no lost numbers. BTW: I always use 3-digit device numbers and CU numbers for CTC. YMMV. -- Radoslaw Skorupka Lodz, Poland ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
