On Friday, 10/04/2013 at 11:04 EDT, "Collinson.Shannon" <[email protected]> wrote: > We're running RHEL6.2 and RHEL6.4 on our zlinux servers under z/VM 6.2 and so > far have been handling HA by just clustering servers--yeah, any transaction in > process to a server that's gone down would be whacked in mid-air, but anything > new would route to the "cluster-buddy" that was still up. That's not true > high-availability, though, and won't work for all our potential applications. > Of course, we're looking at SSI which will help out for planned outages, but > we'd also like to be able to do something in the case of a server-crash (i.e. > have some sort of heartbeat-monitor that could pop up/activate an application > on a different server if it noticed something was down). We'll be > investigating Tivoli Systems Automation for Multiplatform, and the Sine Nomine > HAO (High Availability Option), plus I intend to see if the RHEL HA add-on is > compatible with zSeries, but I'm wondering if there's any other good products > to explore (as well as anyone's experiences with the above products). We did a > cursory look at LinuxHA, but unfortunately our management is not keen on using > freeware, even though price will definitely be a consideration in whatever we > decide on. > > Any comments from those in the field actually exploiting HA for zSeries at > their shops?
You are asking the right questions, but recognize that there is no single HA management solution. Real HA is more than just workload distribution. You have to protect yourself from outages of networks, servers, storage, and the components that connect them together and make them go (adapters, cables, power supplies, etc.) within a single site/campus. DR is a twist on HA that drives it to the next level, achieving the same purpose, but across longer distances and with a higher tolerance for a service outage. As others have noted, a good HA solutions can be leveraged for planned outages, too. Networks and servers are fairly straightforward and well understood (bonding solutions, app clusters, IP moves, "another LPAR", "another CPC"). I find clients who get all that done and then I discover that they have a single storage controller. They might be replicating to the DR site, but that doesn't help them if they lose the local storage frame. If you have z/OS, then you need to look at GDPS, even if only for I/O hyperswap capability. Alan Altmark Senior Managing z/VM and Linux Consultant IBM System Lab Services and Training ibm.com/systems/services/labservices office: 607.429.3323 mobile; 607.321.7556 [email protected] IBM Endicott ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] 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/
