On Thu, 3 Apr 2025 18:19:16 +0000, Allan Staller <[email protected]> wrote:
>OSPF has nothing to do with GDPS per se. > >A GDPS Hyperswap will bring up the cecondary site images, exactly as they >looked in the primary site. Hyperswap does not bring up a secondary image. Hyperswap diverts all the DASD I/O to a set of locally mirrored "secondaries" (fvvo "locally"). GDPS is *also* capable of bringing up and down LPARs, across CPCs, and can automate a site swap. It involves (smoke and) mirrors, but not hyperswap, as (1) they are asynchronous copies, and (2) different LPARs.. >.... > >OSPF will learn the new network connections dynamically without manual >intervention. If you don't have OSPF, that's fine, but you have to engage the network folks to change the location of your VIPA subnets, just as they might do for other parts of your infrastructure in other failure scenarios. It's not rocket science for the network team. BAU. Don't sweat the small stuff. Having OSPF lets you have your VIPAs hanging off of different subnets (A and B). When in site A, your physical IP configuration is configured for subnet A. When in site B, it comes up with a different configuration connecting subnet B instead, so all routes are via subnet B. That oversimplified, of course, but it lets z/OS control the network routing for its VIPAs. For some people (computers don't care), there is value in z/OS being the captain of the ship, but if you're not using OSPF today, that's not a given, and the world is a different place than it was in the 1990s. Much depends on your network team. They must cooperate with you in order for you to use OSPF. If they don't want you to use it, it's their call. They own the stability of the network. And while GDPS is capable of swapping sites automatically when there's an outage, I've never seen a customer do it. They all have selected manual failover, letting the human make the decision, and allowing time to declare the emergency and engage the other parts of the infrastructure in the DR event.. "ALERT! ALERT! ALERT! I tell you three times: All personnel to battle stations! DR failover in T-120 seconds. The ability to stop the failover will end at T-30 seconds. [strobe lights...steam venting...wailing alarms...you know the drill.] Prepare to turn your keys, ladies and gentlemen, ....in 3....2....1....turn!") Alan Altmark IBM z/VM Development ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
