Understanding the Cost Estimation service

The Cost Estimation backend service estimates the costs, forecasts, and migration simulation results that are shown on the TrueSight Cloud Cost Control user interface.

For more information, see the following sections:

Overview

The Cost Estimation service uses the following input parameters to estimate the aggregated costs, trends, and forecasts for the cloud providers. Additionally, for the on-premises infrastructure, the service estimates the costs per instance based on the specified cost definition.:

For on-premises infrastructure:
- Costs of instances in your data center. For example, costs for compute, storage, and networking. You can define these costs on the Cost Settings page. For more information, see Defining-and-managing-the-on-premises-costs.
- Configuration metrics for instances in your data center. For example, number of CPU cores, total real memory, and storage size. These metrics are collected by the ETL modules and stored in the TrueSight Capacity Optimization data warehouse. For example, the VMware - vCenter Extractor Service ETL module collects metrics for the VMware systems. For more information, see Collecting-data.

For public clouds: Detailed costs per line-item for the public cloud providers that are collected by the following ETLs.
- Amazon Web Services - AWS Cost and Usage Extractor
- Microsoft-Azure-Cost-and-Usage-Extractor

The Cost Estimation service runs multiple times a day. If any of the input parameters changes, the service runs again to re-evaluate the costs.

If you change the cost definition of your on-premises instances and want to apply the changes to the past data, you can trigger a manual refresh from the Cost Settings page. The Cost Estimation service uses the recovered data to re-evaluate the costs and the updated data is displayed in the Cloud Cost Control. For more information, see Manually-recovering-historical-costs.

The Cost Settings page displays the date and time when the data was last refreshed in the Cloud Cost Control:

For example, if the ETL imports the AWS data on March 3, 2017 at 8:00 A.M. and the Cost Estimation service processes the data at 9:00 A.M. on the same day, then the date and time displayed are: 03/03/2017 8:00 A.M. If the ETL imports fresh data at 8:00 P.M on the same day but the service has not yet processed it, then the date and time will continue to show the earlier information until the new data is processed.

Cost forecasts

The Cloud Estimation service estimates the cost forecasts for six months in the future.

The forecasted costs are based on the following considerations:

For compute services, the Cost Estimation service applies linear regression. The algorithm adapts to a regime change seen in the recent cost series. A regime change indicates a state when the general trend of the series shows a considerable increase or decrease that continues for some time.
For other services, the major cost contributors such as storage, network, and so on, are considered to estimate the forecasted costs.
For reserved instances, forecasting is based on the average monthly historical costs.
Taxes are applied based on the percentage of the total cost of service. Taxes are proportional to the overall service cost in the previous month. For example, consider that your total cloud expenses were 50K last month along with 1K of taxes. If the prediction for the cloud cost for the next month shows a growth to 100K, then the taxes will be 2K.

Note

Cost forecasts are predictions. They are based on historical cost data. The Cost Estimation service applies algorithms to this data to extrapolate a modeled behavior and to forecast future costs.

The forecasted costs might differ from your actual charges for each month. Here are some of the reasons for the difference:

Start of an extraordinary number of new resources or services, which were not observed in the past periods.
Termination of an extraordinary number of new resources or services, which were not observed in the past periods.
An extraordinary increase or decrease in resource usage, which were not observed in the past periods.

The accuracy of the cost forecasts improves as the amount of historical data increases.

Business service costs

The Cost Estimation service estimates the cost per business service as follows:

For a public cloud	For on-premises infrastructure
Sum of costs of all items in the bill with tag name or key (Service or the user-defined tag name for business service).	Sum of costs of the systems and resources in a business service or in all technical services within a business service. Information The cost of a resource is associated with only one business service even if the resource is assigned to multiple business services. The following screenshot shows a business service hierarchy: Cost of Data Center Services business service = (Cost of phx-jiraprd-02) + (Cost of phx-sqldb-33)

For a public cloud

For on-premises infrastructure

Sum of costs of all items in the bill with tag name or key (Service or the user-defined tag name for business service).

Sum of costs of the systems and resources in a business service or in all technical services within a business service.

Information

The cost of a resource is associated with only one business service even if the resource is assigned to multiple business services.

The following screenshot shows a business service hierarchy:

Cost of Data Center Services business service = (Cost of phx-jiraprd-02) + (Cost of phx-sqldb-33)

Note

Bill items without the business service tag are not considered in the cost calculation for any business service. For example, taxes.

For more information about how data can be collected for a business service, see Collecting-business-service-data.

Where to go from here

Analyzing-and-forecasting-multi-cloud-costs