The cost per MSU (or MIPS) is not steady. At least not for sub-capacity pricing. As the number of MSUs or MIPS you consume increases, the cost per/MSU (MIPS) goes down. This means that each MSU (or MIPS) saved will be the cheapest you are paying for. This can be verified by reviewing the IBM Z Software Pricing Reference Guide.

The price per MSU is based on a multi-level system unique to the sub-capacity pricing metric your organization has agreed to. IBM offers multiple sub-capacity metrics, including Advanced Workload License Charges (AWLC), Country Multiplex License Charges (CMLC), Workload License Charges (WLC), Advanced Entry Workload License Charges (AEWLC), Entry Workload License Charges (EWLC), Midrange Workload License Charges (MWLC), and System z New Application License Charges (zNALC).

There are specific requirements and prerequisites for each of these, and, indeed, you may have negotiated full capacity charges for one of these metrics. But it is not the intent of this article to explain all of these pricing metrics in detail. Suffice it to say that if your organization has agreed to one of these, then the following information can apply to your site.

With sub-capacity pricing[1], your eligible Monthly License Charge (MLC) products will have variable pricing. IBM publishes a price structure for each of the pricing metrics mentioned earlier. Each product will have a price based on the level of MSUs consumed. For example, let’s consider AWLC, which has nine pricing levels as follows:

- Base AWLC: 3 MSUs
- Level 0: 4 – 45 MSUs
- Level 1: 46 – 175 MSUs
- Level 2: 176 – 315 MSUs
- Level 3: 316 – 575 MSUs
- Level 4: 576 – 875 MSUs
- Level 5: 876 – 1315 MSUs
- Level 6: 1316 – 1975 MSUs
- Level 7: 1976+ MSUs

Each level is charged at a different price per MSU, with the highest price per MSU at the lower level. Lower rates ($/MSU) occur at higher levels, meaning that as you ramp up usage, the cost per MSU declines.

In other words, as the title of this piece says: the last MSU is the cheapest.

**Let’s Walk Through an Example**

Let’s assume that you are looking at the peak of 47 MSUs consumed during the month. In this case, we have consumed across three levels: the Base Level, Level 0, and Level 1. So, to determine the amount billed, we need to break up the MSUs across the three levels. The first 3 MSUs are charged at the base rate, the next 42 MSUs are charged at the Level 0 rate, and the final 2 MSUs are charged at the Level 1 rate, as such:

(BaseRate * 3) + (Level0Rate * 42) + (Level1Rate * 2)

The exact pricing of IBM products is not published, so you will have to get this information from your contracts or your IBM representative. If we assume the Base Level is 100, and each level offers a 2% discount, then Level 0 has a rate of 98, and Level 1 has a rate of 96. Plugging these into our formula, we get:

(100 * 3) + (98 * 42) + (96 * 2) = 200 + 4116 + 192 = 4608

Now assume we can implement some tuning efforts that reduce the peak from 47 MSUs to 42 MSUs. This means that our savings would be:

(Level1Rate * 2) + (Level0Rate * 3)

Plugging in our values, we see that we saved:

(96 * 2) + (98 * 3) = 192 + 294 = 486

The rate charged for Level 0 is lower than that for Level 1, so those first 2 saved MSUs have a lower cost-savings than the remaining 3.

This can confuse people who do not understand this pricing structure. For example, the inaccurate assumption may be made that if we were paying 4608 for 47 MSUs and we saved 5 MSUs, then the expectation could be that we save 42/47 of the cost. That is, 4608 / 47 = 98.04, which is assumed to be the cost per MSU. But as we saw, it isn’t… the cost per MSU is variable! In our specific case, the average cost per MSU saved was 97.2 (or 486 / 5).

**The Bottom Line**

Other pricing metrics have different levels and cut-off MSU ratings than the example we walked through (for AWLC). But they are all the same with regard to the price decreasing as the number of MSUs consumed increases. And that means that they all result in the last MSU being the cheapest one.

Of course, none of this is meant to imply that you shouldn’t tune your workload to reduce cost, just that you need to be aware that each MSU saved will not result in an even savings. You cannot just say: “Well, I went from 120 MSUs to 119 MSUs, and I paid $200,000 last month at 120 MSUs, so I should now pay (200000 * 119) / 120.”

No… your savings will be less than that. Understanding how pricing works can help you to accurately predict the impact of cost-saving measures.

** [1] Note: **The non-sub-capacity eligible products will have a single, non-variable charge per product, typically based on the overall capacity of the mainframe (or LPARs) on which it runs(depending on the product, vendor, and contract). So, keep in mind that our discussion here applies to products eligible for sub-capacity pricing only.

© 2022 Craig S. Mullins

#### Find Out More

Learn more about zIIP.