3 ms·
Good you found some. And "somewhat developing?" It's their main, mission-critical platform: http://www8.hp.com/us/en/products/servers/integrity/nonstop/index.h
by nickpsecurity 10y ago
Good you found some. And "somewhat developing?" It's their main, mission-critical platform:
http://www8.hp.com/us/en/products/servers/integrity/nonstop/index.html http://www8.hp.com/us/en/products/servers/integrity/nonstop/...
It's why they ditched OpenVMS. Both advertised similar things for similar markets and hence were competing products. Now, HP doesn't necessarily update NonStop fast as with any legacy acquisition they milk for cash. They are updating and promoting it, though.
Let me look at the manuals. I recall the OS was called Guardian. Found it! Here you go buddy:
http://h20628.www2.hp.com/km-ext/kmcsdirect/emr_na-c02543407-11.pdf http://h20628.www2.hp.com/km-ext/kmcsdirect/emr_na-c02543407...
- kragen 10y agoI think HP's main platform these days is Microsoft Windows on amd64. But yeah, NonStop seems a lot more alive than Itanic, HP-3000, Alpha, VMS, Domain/OS, the HP 9000, PA-RISC, PDP-11, the PalmPilot, the VAX, the Convex, webOS, maybe even HP-UX. In that context, maybe you can see how NonStop's remaining vitality surprised me? And yeah, the Guardian Programmer's Guide you link there was the thing I read the most of last night. I don't suppose you know of any material covering the details of passive failover in process pairs? Because the current version of the Guardian Programmer's Guide pretty much only covers active failover, saying that passive failover isn't compatible with heap allocation.
- nickpsecurity 10y agoLol. Yeah, their track record is only slightly worse than Betamax at this point. HP 9000 or 3000, whichever did high-level machine, was actually kind of neat. Destined to die at that point, though. "I don't suppose you know of any material covering the details of passive failover in process pairs?" Not on-hand. Not compatible with heap allocation make me think it's for reactive stuff like real-time control. Working soon but might try to look it up next day in whatever I downloaded before.
- nickpsecurity 10y ago" Because the current version of the Guardian Programmer's Guide pretty much only covers active failover, saying that passive failover isn't compatible with heap allocation." Just did a quick search to find you an answer. The materials kind of conflate general stuff about backups/redundancy with NonStop-specific stuff. Makes it confusion. Might have hit paydirt here, though: https://h20566.www2.hpe.com/hpsc/doc/public/display?sp4ts.oid=4201311&docId=emr_na-c02128402&docLocale=en_US https://h20566.www2.hpe.com/hpsc/doc/public/display?sp4ts.oi... On page 185 about process pairs, it says this: "Passive backup, in which the backup process passively waits to be activated in the event the primary process fails." "Active backup, in which the backup process actively updates its state while listening for system messages." I could be reading too much into my skimming. However, the initial description looks to me like the passive model doesn't run at all unless a failure already occurred. Whereas the active one constantly gets relevant state from primary. That means passive is inherently stateless in operation. Applications using heap are stateful. So, they need a stateful/active, backup component. I see this playing out in practice where (example) HTTP servers would use passives and databases would use actives. Availability of passive for such needs is probably a cost/performance optimization. Note: Thanks to "spitfire" in another thread for giving me the link below that led to the availability guide. Great they kept the reports up. http://www.hpl.hp.com/techreports/tandem/ http://www.hpl.hp.com/techreports/tandem/ Note 2: I'm keeping that availability guide as it might have nice tips on the topic usable for OSS solutions. I've reposted the OpenVMS Availability Guide a few times because it's still relevant if you swap the proprietary for OSS stuff.