Getting to the Bottom of Unexpected Cloud Cost Spikes

We often work with customers to figure out where cloud cost spikes come from, how to identify waste in cloud infrastructures at scale and why cloud spend increases month-to-month. Let’s take a look at some approaches you can use to get the to bottom of these kinds of questions.

We often work with customers to help answer questions like:

  • “What is this big spike in my cloud costs?”
  • “Where did I waste AWS Reserved Instance hours last month?”
  • “Why are my Usage Hours climbing?”

Let’s take a look at some approaches you can use to get the to bottom of these kinds of questions, using Cloudability Dashboards and Reports.

How it works

Dashboards provide a great way to see an aggregate view across all of your accounts, for monitoring cloud cost and utilization trends.

A lot of times you’ll see interesting — and sometimes unexpected — spikes on your dashboard:

cldy emerge 2016 04 screenshot 1 2710 - Getting to the Bottom of Unexpected Cloud Cost Spikes - Apptio

If you click on one of these anomalies on your dashboard, you’ll get an interactive tooltip that lets you drill down into the data behind the charts for further analysis using our analytics tools:

cldy emerge 2016 04 dashboard drilldown ex 2706 - Getting to the Bottom of Unexpected Cloud Cost Spikes - Apptio

The resulting drilldown report shows that the spike was your monthly AWS Support Charge (these appear on the last day of the month):

cldy emerge 2016 04 screenshot 3 2711 - Getting to the Bottom of Unexpected Cloud Cost Spikes - Apptio

To guide your exploration of the data point in question, you’ll notice above that we included additional dimensions in the corresponding report to make it easier to unpack what caused the changes you saw on your dashboard.

  • For Cost and Usage charts (coming from vendor billing files), we append Product Name, Transaction Type, and Usage Type to your drilldown report.
  • For Utilization charts (coming from vendor APIs like CloudWatch), we append Instance Type, Availability Zone, and Instance ID so you can zero in on the situation.

These fields help you quickly see what was responsible for the spike you were looking at. The Transaction Type field is a great example of how Cloudability enriches the raw data from vendors like AWS and Azure, by adding user-friendly fields to help you better understand the data. It answers questions like, was that cost a one-time charge? A recurring monthly fee? A credit?

Learn more about the Cloudability-created “Transaction Type” field.

Taking the data even further

Once you’ve drilled down from a dashboard to a report, our Success Team also offers some recommendations on how to further analyze the data.

Identifying who’s responsible for cloud costs

If you want to know who was responsible for the cost, there are a few ways to approach it. If your organization manages cost allocation via linked accounts, you can click the “Customize Report” button to add the Account Name dimension to the report. If you have tags setup that indicate resource ownership (we often see ‘Team’ or ‘Developer’ tags), you can also add that tag to the report to see who is responsible for the costs.

Digging into individual line items

If you want to get more information on individual line items, you can add the Item Description dimension, which spells out further details. For example, if you had clicked on the spike on the right side of the chart (the biggest one) in the example above, you would have drilled down to see if was your monthly recurring Reservation charges. To see how many hours you’ve used this month, and what the hourly rate is, you could add Item Description to see the full story:

cldy emerge 2016 04 screenshot 4 2712 - Getting to the Bottom of Unexpected Cloud Cost Spikes - Apptio

Also note that once you’ve drilled down from a dashboard to a report, you can leverage the capabilities of our analytics tools. Further slice and dice the data with live filters (click any value in the table to filter down the results), setup email alerts if it ever happens again, share your findings with a colleague, or export a CSV for analysis in other tools.

Creating time period comparisons

Finally, drilldown exploration works really well when you have selected a time-period comparison for your chart (using the date picker you can compare week over week, month over month, etc.). The pinned tooltip lets you compare differences in specific data points side-by-side on your dashboards:

cldy emerge 2016 04 screenshot 5 2713 - Getting to the Bottom of Unexpected Cloud Cost Spikes - Apptio

And then when you drill down to the report for more details, you can see exactly where the major differences are for the date you clicked:

cldy emerge 2016 04 screenshot 6 2714 - Getting to the Bottom of Unexpected Cloud Cost Spikes - Apptio

Give this tutorial a try – go and click on a spike on one of your dashboards!

Article Contents

Categories

Tags

Additional Resources