But according to the datasheets, upgrading, say, an H06 to an H13 "requires planned down time", so if you started small and then want to grow, the only feasible (non-down-time) upgrade path is to buy a 2nd mainframe, which, as you point out "won't scale painlessly".
With a COTS based system, you work with a cluster configuration from the start (without requiring additional licenses), and have a rather more granular and disruption-free upgrade path. And because it is designed from the ground up as a cluster, it is designed to be maintainable WHILE WORKING. Replacing any hardware component of the cluster (ECC memory, CPU, I/O board, main board, network component, rack, ...) can be done while the system is running. So there isn't really any *functional* advantage to using a mainframe. The question is whether you want to be running a cluster of, say, 2 - 5 mainframes, or, say, 10 - 500 COTS boxen. I.e. "what do you want to spend your money on". And no, you should not then have a bunch of sysadmins running around manually managing those 500 COTS boxen. That's supposed to be automated... WFK On 05/25/17 16:22, John Campbell wrote: > As I recall from Appendix A of the "Linux for S/390" redbook, the S/390 > (and, likely, zSeries) is designed to be maintainable WHILE WORKING. > > The multi-dimensional ECC memory allows a memory card to be replaced WHILE > the system is running. Likewise, power supplies the CPs. > > I have to agree that the "second" zSeries box won't scale painlessly; The > work to load balance would NOT be fun (and the second box has its own > issues w/r/t the management team, too). > > I recall, when dealing with the idea of putting an S/390 into a Universal > Server Farm in Secaucus, NJ (I had some fun helping define the various > networks as this predated the "hyperchannel" within the BFI ("Big Iron") as > part of this USF integration) when it was killed for non-technical reasons. > > -soup > > On Thu, May 25, 2017 at 8:41 AM, Philipp Kern <pk...@debian.org> wrote: > >> On 24.05.2017 00:03, John Campbell wrote: >>> Cool... >>> >>> Though the real key is that the mainframe is designed for something at or >>> beyond five 9s (99.999%) uptime. >>> >>> [HUMOR] >>> Heard from a Tandem guy: "Your application, as critical as it is, is on >> a >>> nine 5s (55.5555555%) platform." >>> [/HUMOR] >> >> Mostly you trade complexity in hardware with complexity in software. >> Mainframes do not scale limitless either, so you trade being able to >> grow your service by adding hardware with doing it within the boundaries >> of a sysplex. >> >> Your first statement is also imprecise. It's designed for five 9s >> excluding scheduled downtime. If you use the fact that hardware is >> unrealiable (after subtracting your grossly overstated unreliability) to >> your advantage, you end up with a system where any component can fail >> and it doesn't matter. You win. >> >> Again, it then comes down to the trade-off question if you're willing to >> pay for the smart software and the smart brains to maintain it rather >> than paying IBM to provide service for the mainframe. >> >> Kind regards >> Philipp Kern >> >> ---------------------------------------------------------------------- >> For LINUX-390 subscribe / signoff / archive access instructions, >> send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or >> visit >> http://www.marist.edu/htbin/wlvindex?LINUX-390 >> ---------------------------------------------------------------------- >> For more information on Linux on System z, visit >> http://wiki.linuxvm.org/ >> > > > > -- > John R. Campbell Speaker to Machines souperb at gmail dot > com > MacOS X proved it was easier to make Unix user-friendly than to fix Windows > "It doesn't matter how well-crafted a system is to eliminate errors; > Regardless > of any and all checks and balances in place, all systems will fail because, > somewhere, there is meat in the loop." - me > > ---------------------------------------------------------------------- > For LINUX-390 subscribe / signoff / archive access instructions, > send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit > http://www.marist.edu/htbin/wlvindex?LINUX-390 > ---------------------------------------------------------------------- > For more information on Linux on System z, visit > http://wiki.linuxvm.org/ > ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/