Apptio for All

Apptio for All

A place for Apptio product users to learn, connect, share and grow together.

 View Only
Back to Library

[++] TBMA Challenge: Assess impact of wrong allocation strategy 

Tue September 19, 2017 09:16 AM

TBMA challenges are designed to help sharpen your TBM and Apptio skills. See title for rating: [+] Easy   [++] Moderate   [+++] Challenging

 

OK TBMAs who never back down from a challenge. You know how you are  We've got a scenario. If you've got the right answer, we'll update your profile with the Apptio U Star award  (worth 30 points!)

 

We'll follow up Oct 4 with the answer and badge winners. Make sure to get in the game before then by reviewing Premier CSM @Chris Davidson's scenario below and clicking the button to submit your answer.

 

 

In my cost transparency project's Cost model, this month I'm allocating $200,000 from IT Resource Towers object (where IT Resource Tower Name = "Data Centers") to Data Centers object.

 

I have 8 data centers.

Originally I was just evenly spreading the money: $200,000 / 8 = $25,000 per data center (DC).

 

Then I read Jenny's post and realized there were better ways to estimate per-DC costs.

Data centers are full of tracked metrics: square footage, occupied volume, rackmount units (RU), power, cooling, and more.

 

I'm torn between a space-based and a power-based allocation weighting strategy.

I could go with square footage. Or power consumption. Or a hybrid of both. Or something entirely different.

I could instead set up multiple allocations to accommodate different incoming cost types.

 

My total monthly tracked IT cost is $4,000,000 by the way.

 

Before I choose a weighting strategy for the $200,000/mo DC cost, my more pressing issue is:

 

What is the impact of choosing incorrectly?

 

More specifically: How can I calculate worst-case scenario effects (if my weighting strategy is absolutely the opposite of reality) or sample average error effects (if my weighting strategy is perhaps close to but not 100% reflective of reality)?

I need to be able to show my manager quantitative evidence of why I should or should not spend 2-4 hours researching and improving this specific model allocation.





#ApptioforAll

Statistics
0 Favorited
0 Views
0 Files
0 Shares
0 Downloads

Comments

Oliver Smith
Mon October 16, 2017 05:03 AM

Thanks Heather!


#ApptioforAll

Apptio Community Member
Mon October 09, 2017 04:45 PM

Congratulations!


#ApptioforAll

Julie Batty
Fri October 06, 2017 04:29 AM

Well done @Oliver Smith and @Edgaras Krusas and thank you @Chris Davidson was a fun challenge to take part in & can't wait for the next one


#ApptioforAll

Debbie Hagen
Thu October 05, 2017 03:40 PM

How fun!

Thanks everyone for playing and Congratulations to @Oliver Smith and @Edgaras Krusas for earning the Apptio U star award!

Can't wait to see more of these and for more people to get involved.


#ApptioforAll

Doug Copeland
Thu October 05, 2017 07:52 AM

Really cool idea Chris.  Sorry I missed this.  Hope to see more stuff like this...really cool.


#ApptioforAll

Oliver Smith
Thu October 05, 2017 03:38 AM

Woohoo! Thanks very much Chris, thoroughly enjoyed the challenge and I'm definitely looking forward to the next one  (No pressure to bring it out quickly or anything )

@Sasha Dunford


#ApptioforAll

Chris Davidson
Wed October 04, 2017 05:00 PM

Congratulations @Oliver Smithand @Edgaras Krusas for earning the Apptio U Star award with your responses.

 

Several respondents gave good suggestions for how to allocate data center costs from IT Resource Towers (ITRT) object to Data Centers object in the Cost model. Your opinions seem split on whether to use space-based, power-based, or a hybrid calculation to fairly split cost between multiple data centers.

 

A few of you correctly identified the worst case scenario: allocating the full $200,000 to one data center, but in reality another data center actually incurred the full $200,000 cost. Obviously this is a highly unrealistic scenario, but let's consider it anyway.

 

Say we allocate $200,000 to DC1 and $0 to DC2-8, but in reality DC1-7 incur $0 and DC8 incurs $200,000.

We're misallocating the full $200,000 (5% of total IT spend).

 

Say instead we use an even spread: $25,000 to DC1-8, but in reality DC1 incurs $200,000 and DC2-8 incur $0.

We're underallocating DC1 by $175,000 and overallocating DC2-8 by $25,000 each.

We're misallocating $175,000 (4.375% of total).

 

But it's more realistic to assume data centers are somewhat closer to each other in total (real) cost.

Even though we can't calculate the exact difference in cost per data center, we can still reasonably estimate the range.

 

If our data centers are similar (by order of magnitude) in managed space and power usage, then the cost difference between each data center is probably closer to $5,000 (0.125% total) to $50,000 (1.25% total), not $200,000 (worst case).

 

Is it worth the time (hours? days, perhaps?) to study data center allocation strategies and potentially improve the cost allocation from ITRT object to Data Centers object? Is 0.125% - 1.25% total IT spend crucial for our data center reports (and presumably our applications and services which eventually receive cost from our Data Centers object)? I'll leave that decision up to you.

 

When you're deciding on priorities for project improvement, include impact estimates which take reasonable average expected error into consideration.


#ApptioforAll