Rob wrote:
What it didn't address is what about OS upgrades, ptf installation
and that ilk? The fact that you have redundant boxes but they
both rely upon the same disk makes both boxes out of commission
when you are upgrading the OS on that disk.
It looks like you missed something with PowerHA SystemMirror.
It is based on using Independent ASP for applications and data.
There are two servers each one with its own SYSBASE. And there also two
disks spaces for the IASP (well, you may have only one disk space for
IASP if using LUN level switching or switched disks, but the principles
for upgrading are the same), each one connected to one server.
Replication occurs at IASP disk system level, either with geographic
mirroring (done, in this case by IBM i), or with disk system based
replication (metro mirroring done by the disk systems). So that
both SYSBASE are running and available.
When you upgrade a server from a release to another, it applies to
SYSBASE. It is the same for applying PTFs. So, in both cases, you may
want to apply PTFs or upgrade the release on the preferred backup
system, then switch over, run production on the preferred backup
system, apply PTFs or upgrade the release on the preferred production
system which is the actual backup system, then switch back to the
preferred roles.
There is a short description of these procedures in chapter 15.9
Upgrading IBM i and PowerHA release of last redbook about PowerHA
http://www.redbooks.ibm.com/abstracts/sg247994.html
It looks to me to be close to what you are doing with a software based
replication tool.
midrange.com
