AIX, PowerVM, and EqualLogic

We're purchasing a Dell EqualLogic PS4000X.  The primary purpose of the EQL is to support a 3-node VMware vSphere Enterprise cluster.

We have an AIX 5.3 - based app that we'd like to provide some protection for as well.  The app is currently running on an older pSeries box.  One idea being floated is to purchase two small but new AIX boxes running PowerVM and build something like a 'cluster' from them with the existing AIX 5.3 OS and app virtualized and sitting on the SAN.

Is this kind of thing possible with PowerVM?  It seems like it would be but I have found precious little on the subject.  Also, is PowerVM a separate product or does it come with new pSeries machines?  If it's a separate item, how much does it cost?  We're just trying to get a sense of the scope and feasibility of this particular idea.  Any help will be appreciated.
Who is Participating?
AIX version 5.3 is supported as a guest VM on the Power6 p520, no problem, but you need the PowerVM feature coming with AIX 6 to be able to virtualize on a Power6 system.
There is no "failover" solution with AIX yet, as is with VMWare.
The "High Availability" solution is called PowerHA, and is a bit expensive, as I wrote.
There is something similar to VMotion included in the PowerVM Enterprise Edition, called "Live Partition Mobility", but you need the "source" hypervisor ("IVM" in PowerVM terms) up and running for it to work. So if there is a total hardware failure, LPM is useless.
Of course you can configure SAN disks to be available at two different VMs, each equipped with a basic AIX installation, to manually import the disks into the running system when the other system fails, to then start the application there. Additionally you must take over the relevant IP address(es) manually, of course. Given all application data reside on the "shared" disks, this is indeed a viable solution.
As for PowerVM in general - please visit this link, to get access to almost all information about PowerVM:
Please don't get nervous in view of those lots of material. Start with the "Introduction" (strange idea!) and take your time ...

of course you can do all you described above wit AIX and PowerVM.

If it's really for a rather small system I'd recommend the Power520 Express (8203-E4xx),
which is available with 1- to 4-core (4.2 - 4.7 GHz) Power6 processors, 2 to 8 GB memory, dual port onboard Ethernet, 2 x 146 GB SAS HD. SAN HBAs are optional, but available.

The AIX/PowerVM express edition license (with only 3 configurable partitions, shared processor pool and the Integrated Virtualization Manager (IVM)) is roundabout $$ 300,- to 1200,- (depending on the model chosen) with a mandatory 1-year maintenance agreement at ~ $$ 250,- to ~ 900,-
I'd recommend the standard edition, with micropartitioning (10 partitions), to configure partitions in steps of 0.1 processing units. The license is about $$ 500,- to 2000,- (see above), and the 1-year maint is ~ $$ 350,- to 1400,-
There is an enterprise edition available, including "Live Partition mobility". Calculate a factor of ~ 1.5 against the standard edition cost.

Your existing AIX 5.3 OS does most probably not include virtualization. You can only use it to run one non-virtualized "full system partition" per server.

What do you have in mind when saying "cluster"? Of course you can buy a High Availability Cluster Software (PowerHA) from IBM, but this software is rather expensive (calculate multiples of the AIX license cost!) IBM have kind of "customer-specific" pricing models for this, so I can't tell you exactly.

So where it's a perfect idea to use Power Servers, AIX and PowerVM, I'd suggest assessing your needs for a sophisticate HA solution like PowerHA very carefully!

Don't hesitate to ask for more/more detailed info. I'm here to assist you.


CeleritasPrimeAuthor Commented:

Thanks very much for your excellent response.

To tell the truth, I am an x86 guy trying to understand the AIX and PowerVM world.  

The application we're running in AIX 5.3 will not support newer versions of AIX, so I'd like to virtualize that OS instance as close to 'as-is' as possible.

What I mean by 'cluster' is functionality similar to VMware's High Availability and/or Failover features.  In the VMware world, if a host fails then the VM is either restarted on the other host (High Availability) or can actually have a shadow running on the surviving host and go along as if nothing happened (Failover).  If we can't do that, then we do at least need to be able to have an easy-to-follow way of getting the other host to process the VM.

Your response is very very helpful, and hopefully I will be able to get my head around PowerVM and how it is or is not like what I'm used to with VMware.

Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.