Image Management for IBM i
One of the many announcements IBM made last week might have slipped past you but is a key step on the roadmap to “Cloud” capabilities for IBM i. It is called “Image Management” in the announcements (it's buried a bit but you can find it on page 9), but also might be thought of as “Virtual Machine Deployment.”
A little education is necessary, I think.
Suppose you have an application, and you want to provide that application on several partitions. The traditional method of accomplishing this is to install every piece of software you need, from the operating system on up through the application, on each of the partitions.
When the functions announced last week become generally available, there will be another option.
PowerVM has a component called VMControl. “VM” can stand for a number of things, but in this blog, let’s use it to mean “Virtual Machine.” You can think of a virtual machine as the partition, and it needs to be loaded with software to make it useful.
As seen in the first figure, an “image” is the set of software needed to run a specific environment, for example, an application. It includes the application itself, as well as any other software required for the application to run. I call this the “application platform.”
As I already implied, though, that bottom rectangle comprises several pieces of software. Without getting too granular, the second figure is one way to think of the “Application Platform.” It includes the operating system, the fixes for the OS and the middleware required.
The basic concept for “Image Management” is this: Install one base partition with everything you need, then use VMControl to create a “base image.” When you want to roll out another partition with that same application, just create the partition based on the base image.
In the third figure, I have used the base image to create three partitions.
Notice that, using the image management software, I was able to create these environments without having to install the software in the image on each subsequent partition. This is much more efficient, and less error-prone, than having to install each of the pieces of software on each partition.
There are a few things to customize for each partition, of course. The VMControl support allows the operating system to do that customization as part of starting the new image on the partition, and that’s why we needed to add support to IBM i 7.1, in Technology Refresh 3, to enable PowerVM’s functions to work with IBM i.
I am particularly excited about this feature, because it is one of the last “basic cloud” capabilities that an operating system needs to have if it is going to be used to implement pure cloud solutions. I use the following chart in many of my presentations to explain that IBM i has been working for quite a while on “cloud enablement” technology. As you can see, this capability, requiring POWER7, IBM i 7.1 TR 3, and the new PowerVM capability, brings us very close to the end of that base roadmap.
Now, image management can be used, as I’ve described it, to create many partitions with the same base application image. But there are other uses for it, too. You could create a master image with only the OS and a set of fixes, then create a new partition easily whenever you wanted. A simple test of a new piece of software would be much easier if getting a new partition up and running did not require a complete installation of the OS, because the base image could be deployed in minutes. Checkpoints could be created. Quality Assurance partitions can be rolled out quickly. The flexibility of this sort of technology is impressive and gets more impressive with advancements we’ll make over the next couple of years.
Twitter: #ibmi #powervm @Steve_Will_IBMi