Published: January 08 2008, 09:04 AM | 1 Comment(s)
by Reg Harbeck

Imagine, if you will, a luxury sports sedan with a V8 engine, plus an extra row of cylinders, designed especially for long-distance driving, that used almost no fuel.

When driving around town, or accelerating out of a traffic light, or performing other day-to-day tasks, you'd use the V8 and expect normal fuel economy.

But when departing for long business trips or vacations, once you hit the highway the special cylinders would kick in, and you would begin to run almost indefinitely before having to stop for gas.

Now, imagine that this extra row of cylinders was a relatively inexpensive add-on option for your car.

Who wouldn't buy it?

As far as I can figure, that seems to be the idea behind IBM's specialty mainframe engines such as the zIIP (z9 Integrated Information Processor) and zAAP (System z Application Assist Processor). You plug ‘em in, and suddenly, specialized workloads begin to run almost "free" in terms of software-billable CPU time. And as a result, according to IBM, zIIP and zAAP growth is accelerating, with a 110% YTY increase in adoption.

Of course, if such a car engine existed and began to get market buy-in, it would make sense that people's driving habits might start to favor those roadways for which this engine was built. Which would likely lead to the opening of more roadways for this specialty use.

Likewise, some pretty cool things are beginning to happen in this new zIIP-and-zAAP-enabled mainframe world.

I know this because I work for an ISV that has begun to configure some popular "software roadways" to be eligible for these cost-saving engines.

In fact, while there's a whole range of our mainframe software from network, database, resource and performance management to databases that will employ and/or monitor these engines, one of the coolest things we're doing is creating software appliances that use this virtual cost-savings mechanisms.

That's right: software appliances using specialty engines to avoid the costs and limitations of traditional approaches.

According to my mainframe storage colleagues, that's what we're now doing with virtual tape drives and tape encryption. Instead of hardware lock-in on the one hand, or high CPU charges on the other, CA Vtape and CA Tape Encryption are running as virtual as you can get. You might even call it virtue-al.

So I'm thinking, what will they think of next? Will IBM introduce another engine that makes software even cheaper on the mainframe? Will ISV's such as CA introduce more software ways to virtualize our customers' mainframe costs with even more innovative uses for this technology? Or will some even newer virtue be found to benefit the mainframe world?

What do you think will happen next?