>>> On Wed, Jun 25, 2008 at 12:21 AM, in message <[EMAIL PROTECTED]@sbcglobal.net>, Ron Hawkins <[EMAIL PROTECTED]> wrote: -snip- > It gets even more ridiculous when Linux is suddenly an anointed HA OS simply > because it will run on an IBM Mainframe, along with Solaris and pre-RISC > AIX. I have not figured that out yet.
Umm, no. It's not the fact that Linux is running on the mainframe that creates high availability. As with all other operating systems, it's the infrastructure that supports HA, as well as (sometimes) some applications that are HA aware. Heartbeat, various STONITH tools and so on are what accomplish HA, not the OS itself, per se. You can get HA without the application being HA aware, but it's easier if they are. There are HA Linux clusters that have never been anywhere near a mainframe. The same holds true for Solaris, HP-UX, Tru64, AIX, and others. Having said that, having your operating systems on System z hardware makes getting HA levels of uptime more common, but not more doable. While server class midrange hardware can be very good, it's not nearly as good as System z. Which partly explains the huge price differential between them. (The rest is due to other factors that are far less satisfying to contemplate.) Even on System z, however, you still have to design the architecture as though any part of it can fail in the next few minutes. Otherwise you're just burying your head in the sand. Finally, don't compare server operating systems to Windows XP. I no longer have any love for Microsoft and the various incarnations of their Windows desktop operating systems, but they're not comparable to the server editions in the same family. Microsoft's desktop OSes don't have any sort of HA potential built into them, but their server versions do, via clustering, albeit along the lines of Microsoft's vision of what that means. Given a choice, and application availability, I would go with Linux on System z (running on z/VM) clustering versus Parallel Sysplex, for no other reason than the huge cost savings, and relative simplicity. That's a purely economic evaluation, not a technical one. I supported MVS for 20+ years, and have a great admiration for it. IBM's pricing schemes (along with most ISVs) make it difficult to justify economically these days. Mark Post ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
