Keeping your IBM Power Systems highly available

We’ve all heard of the “blue screen of death.” It’s the dreaded crash on a Windows-based computer. The potential loss of everything. If you’ve had the misfortune of seeing one, you know how upsetting it is. To put it mildly, it’s not fun.

Read More: C2150-609: IBM Security Access Manager V9.0 Deployment

An operating system crash of any kind leaves application users dead in the water. So avoiding single points of failure (SPOF) is a high priority. The UNIX® operating system is a potential SPOF, and this is true of all UNIX based operating systems including versions such as AIX®, HP-UX, iOS and all Linux flavors such as Ubuntu, RHEL, Android and SUSE.

IBM’s solution

Fortunately, thirty years ago IBM introduced high availability cluster multi-processing (HACMP) to combat this issue on its AIX RS/6000 systems. To state things in the simplest terms, two or more systems are paired in a cluster. When one system crashes, the other takes over. There is an application outage, but recovery is automated and downtime is primarily dependent on how long the applications themselves take to recover/restart. In the days without storage area networks, IP aliases or virtualized machines, HACMP/PowerHA® was a terrific way to go.

HACMP software is still widely used on the newest IBM Power Systems, having been renamed PowerHA SystemMirror®. But fourteen years ago, IBM introduced Live Partition Mobility (LPM). LPM gave users the ability to move fully virtualized AIX machines between physical hardware servers while the applications continued running. Ultimately, this is a game changer in high availability.

VM Recovery Manager High Availability (VMRMHA)

VMRMHA utilizes LPM simplified remote restart (SRR) technology for high availability. A user utilizing VMRMHA can set up an HA environment not just for selected virtual machines, but every virtual machine on an IBM Power® server. This includes Linux and IBM i virtual machines as well.

One drawback of using VMRMHA versus PowerHA is that with VMRMHA, the operating system must boot on a new physical server after an OS crash, and then the application restarts. In a PowerHA environment, only the application must restart. However, using VMRMHA has an advantage over PowerHA in that there are no “cluster of two AIX servers” that must be maintained or kept in sync.

To summarize:

No “cluster” to maintain.

– plus –

Only one AIX OS to maintain.

– plus –

Only one set of application code to maintain.

– multiplied by –

Number of PowerHA clusters

= equals =

Application High Availability is easier to maintain!

Converting existing PowerHA/HACMP clusters to a VMRMHA strategy is not a trivial task, but in the long term, it’s worth it because of reduced administrative overhead.

Source: ibm.com