Debbie Hagen

How different allocations can affect the numbers in your Models and Reports

Blog Post created by Debbie Hagen Employee on Aug 29, 2018

Hello Apptio Practitioners,

 

One question I'm often asked in Lab Assists and on Ask an Instructor  is "why don't my numbers in Servers and Hypervisors match the guide or the answer key?  What did I do wrong?  

 

Well the short answer is you may not have done anything wrong.  It's typcially related to slightly different allocation strategies since we're not giving you all the answers and are expecting you to figure some things our for yourself as you get farther along in your training.  But the question comes up enough that I want to explain why it happens using an example from the CT Apps and Services project. 

 

Let's say you configure your allocations from Physical Servers to Hypervisors and Servers as follows:

 

 

Some of you may get a result that looks like this:

 

 

 

While some of you may get different numbers.

 

 

 

So why does this happen, especially since in both cases the allocations from Physical Servers to Servers and to Hypervisors are identical?  Well it turns out, that the reason is back a couple steps in our project with the allocation from Data Centers to Physical Servers. You may think "but I've got the right number in physical servers"  and you're right, but sometimes its about the distribution of those dollars. 

 

To explain this, let's take a look at the allocation from Data Centers to Physical Servers that we did earlier in the project.  Let's say you allocate from Data Centers to Physical Servers using Location as your data relationship column... but you don't select any weighting value.  Apptio spreads the dollars from the Data Center line items to the respective Physical Server Line items evenly.  In the screenshot, I've filtered by Raleigh so you can see that every server in Raleigh gets the same $ amount.

 

Now let's change the allocation slightly.  I'll add a weighting factor to my allocation based on the CPU Cores and once again filter for Raleigh.  Notice how when it's weighted by CPU cores, the higher the number in the CPU Core column the more dollars are allocated.  Lower CPU Core numbers result in lower allocations.

 

The point is, regardless of which allocation strategy I use, I get the same number going from Data Centers to  Physical Servers:  

 

The difference is in the spread, and that comes out later when we allocate from Physical Servers to Servers and Hypervisors.  Because the $ allocated to each Physical Server line item are different in the different allocation strategies, the amount to send from specific line items in Physical Servers to Hypervisors or Servers are different, which changes the amounts in those subsequent allocations.

 

We can compare a specific line item.  Below is the project where I have a CPU Core Weighting in my DC to Phys Serv allocation.   I've filtered for HV1000272.  Notice that in my Cost column for that line item in the Source data,  I have $7352 to allocate.l

 

 

Now, if you look at the same line item in the project where I've not used the CPU Core Weighting and just let the dollars be spread evenly we get different results. Notice that we only have $2408 available to allocate from the HV100272 line item in Physical Server Master data to the relevant Hypervisor line item.  Of course this is going to change our results.   

 

 

The point is, different companies use different allocation strategies based on what is important to them. Spreading the dollars evenly, without a weighting is a perfectly valid allocation strategy although not quite as accurate and defensible as one that is based on an attribute of the data or on consumption. Note: If you're trying to fix your project to match the guide, you can do so by adding the CPU Cores weighting to the Data Center to Physical Server allocation.  

 

I hope this lesson on how the weighting factors can change your results has been helpful.  Let me know what you think!

 

Happy TBMing!

Outcomes