What you need to know about blended rates on AWS
When an AWS user has linked accounts within a Consolidated Billing account family, AWS calculates something called a “blended rate” and includes it in the billing reports for that customer. We get a lot of questions from our own customers about what exactly this means and what it should be used for, especially compared to the “unblended rate.” And unfortunately, there’s a lot of misinformation out there to wade through.
While blended rates are sometimes described as an “average” of on-demand and reserved spending across usage types that you can use to determine how effective Reserved Instance coverage is across your linked accounts, the reality is not so simple. In fact, the blended rates that AWS calculates for each of your usage types can be challenging to interpret—and misunderstanding their significance can be costly.
The formula for blended rates
AWS calculates a blended rate and blended cost for each unique type of usage that has been generated by your account during the reporting period (for example, “m3.large, us-east-1a, Linux” would constitute a single usage type). Then, AWS uses those values to calculate your total blended rate and total blended cost across all usage types.
In the past, AWS has used two formulas for calculating the blended rate of a usage type: one for usage that was at least partially covered by Light and Medium Reserved Instances, and one for usage that was at least partially covered by Heavy Reserved Instances. However, now that Light and Medium Reserved Instances are no more, the formula previously used for usage partially covered by Heavy Reserved Instances is used ubiquitously for calculating the blended rate for all usage of a single type—no matter whether that usage is purely on-demand, or includes Partial, All, or No Upfront Reserved Instance coverage.
This formula can be complicated to understand—and now that it’s used across the board, interpreting its results can be tricky. Here’s what it looks like:
The formula effectively divides the on-demand costs of a usage type by all of the hours (reserved and on-demand), without including the hourly costs of Reserved Instances. This means that the blended rate is not a comprehensive representation of your spending; you still have to account for Reserved Instance costs.
We can visualize this with a quick example. Say you’re calculating the blended rate for the usage type “m3.large, us-east-1a, Linux,” in which you accrued two hours of spending at an on-demand rate of $0.14 an hour:
Now, say you’re calculating the blended rate for two hours of on-demand spending, plus an additional hour covered by a Partial Upfront Reserved Instance:
As you can see, the calculation in this example results in an hourly rate lower than that of the first example, in spite of the fact that you’re paying for more instances during that hour. That’s because not all of the costs are included within this value. If you were to wrongly interpret this as your comprehensive rate or base your assessment of your Reserved Instance coverage on this value, you’d be in for quite a shock at the end of the month. That’s because you still have to account for the costs of your Reserved Instances, which are represented elsewhere in your bill.
Accounting for Reserved Instance costs
To find the costs that aren’t accounted for in a usage type’s blended rate, you’ll have to turn to the injected line items, either by looking at your Detailed Billing Report or by using certain filters in your Cloudability reports (which we’ll touch on more in a bit). The injected line items are the lines in the report that account for your monthly Reserved Instance costs, broken out by Reserved Instance class (each usage type has its own corresponding Reserved Instance class). Here’s an example of Reserved Instance spending displayed in an injected line item:
The BlendedCost column in this row contains the monthly installment of your Reserved Instance payment. These monthly sums are distinct from signup fees, which are located in separate items elsewhere in the DBR and aren’t factored into your blended or unblended rates. All Upfront Reserved Instances are paid for entirely in the signup fees, and will therefore not show charges within these lines as their complete costs are accounted for elsewhere. Partial Upfront and No Upfront Reserved Instances will have costs to show here, though. Regardless of which Reserved Instance types you use, this column will contain all of the “missing” costs which aren’t accounted for in the blended rate calculation for that particular usage type.
Although you can easily identify your Reserved Instance costs within the Detailed Billing Report by checking these items, you can also access this data with Cloudability. Simply use the transaction type dimension to identify your recurring charges, and the item description dimension to identify your specific Reserved Instance costs:
It is important to note that these injected line items cannot be tagged. You should therefore be careful not to accidentally exclude them from your reporting if you want to include these costs in your analyses.
The anatomy of total blended cost
When AWS calculates the total blended cost across all of your usage types (which you’ll find listed at the bottom of your Detailed Billing Report), it adds up the blended costs for each unique usage type with the Reserved Instance costs found in the injected line items. This means that your total blended cost will include your reserved spending and will therefore represent your true spending—in fact, it’s what AWS references in determining your total invoiced cost.
However, as we’ve shown, your blended rate and blended cost for a unique usage type will not account for reserved spending unless you factor in the costs listed in the line item, and must therefore be referenced carefully.
Blended vs. unblended
Parallel to the blended rate and blended cost calculations, AWS also lists an unblended rate and unblended cost.
The unblended rate represents how much was charged for a specific instance-hour of a particular usage type. If the hour was covered by a reservation, the unblended rate will indicate as much by listing the reserved rate. If the hour was not covered by a reservation, the unblended rate will list the on-demand rate. The total unblended cost will sum these values across all of your instance-hours to represent your true spending.
In other words, the unblended rate will always represent exactly how much you paid on-demand, and how much you paid reserved, without combining those two rates or values.
Unblended cost and blended both have their applications. The Cost Allocation File and Amazon Invoices both reference blended cost for their data, but Amazon Cost Explorer references unblended cost. You can achieve accuracy in your own reporting by keeping their differences in mind and using the appropriate rate to your needs. For example, you might reference blended cost to see a historic representation of how you’ve been billed by AWS, but reference unblended cost to see the impact of Reserved Instances on your bill.
You can remind yourself of the differences between unblended and blended cost with this table:
Whether calculated using blended rate or unblended rate, the total costs in either case will be equal to each other.
Whether you’re drilling into your blended rates or your unblended rates, Cloudability reporting includes dimensions that have got you covered. Graph your blended rate over time, compare unblended rates across usage types, and more; simply log in or sign up for a free 14-day trial to get started.