Without a doubt, AWS has fundamentally changed how modern enterprises deploy IT infrastructure. Their services are flexible, cost effective, scalable, secure and reliable. And while moving from on-premise data centers to the cloud is, in most cases, the smart move; once there managing your costs becomes much more complex.
On-premise costs are straight forward, enterprises purchase servers and amortize their costs over the expected life. Shared services such as internet access, racks, power and cooling are proportionally allocated to the cost of each server. AWS on the other hand, invoices each usage type separately. For example, if you are running a basic EC2 instance, you will not only be charged for the EC2 box usage but also the data transfer, EBS Storage and associated snapshots. You could end up with as many as 13 line items of cost for a single EC2.
Example: Pricing line items for a single c4.xlarge Linux virtual machine running in the US East Region (Click on image to view larger)
When examining the composition of various workload types the numbers of line items to manage will vary. A traditional VM-based workload may have 50 cost line items for every $1,000 of spend while an agile, cloud-native workload may have as many as 500 per $1,000 and a dynamic workload leveraging spot instances may have upwards of 1,200 per $1,000. This “parts bin” approach to pricing makes the job of cost account challenging.
To address this complexity and enable accurate cost accounting of your cloud costs; we recommend creating a business-relevant financial tagging schema to organize your resources and associated cost line items based on your specific financial accounting structure.
Here are some recommended financial management tags you should consider (Click on image to view larger):
AWS Tagging data integrity is extremely important in ensuring the quality of the information it provides and is directly dependent upon the rigor applied in adopting a systematic and disciplined approach to AWS Tagging.
Financial Management Tagging – Best Practices
- Create a framework or standard for your enterprise that outlines required tag names, tag formatting rules, and governance of tags.
- Tags should be enforced and automated at startup of the resource via Cloud Formation templates or other infrastructure as code tools, such as Terraform, to ensure cost accounting details are captures from time of launch.
- NOTE: Tags are point in time based. If a resource is launched without being tagged and then tagged sometime in the future, all hours the resource ran prior to being tagged will not be included in tag reports in the AWS console.
- Manually creating tags and associated values is strongly discouraged as it leads to miss-tagged and untagged resources and in-accurate cost accounting
- Select all upper case or all lower-case keys and values to avoid discrepancies with capitalization.
- NOTE: “Production” and “production” are considered two different tag names or values.
- Monitor resources with AWS Config Rules and alert for newly created resources that are not tagged
Once your tagging schema is created, automation is in place to tag resources during startup and alerts are set up to ensure tagging is managed, you can accurately to view, track and report your cost and usage using any of your tagging dimensions.
Financial Management Reporting – Best Practices
- Using your tagging schema, group your resources by workload.
- Apply Reserved Instance discounts to the workloads you intended them to be for.
- NOTE: 2nd Watch’s CMP Finance Manager tool converts reserved instances into resources so that you can add them to the workload they were intended for.
- Organize your groups to match your specific multi-level financial reporting structure.
- Managed shared resources
- Create groups for shared resources. If you have resources that are shared across multiple workloads such as a database used my multiple applications or virtual machines with more than one applications running on it, create groups to capture these costs and allocate them proportionally to the applications using them.
- Manage un-taggable resources
- Create a group for un-taggable resources. Some AWS resources are not taggable and should be grouped together and their associated costs proportionally allocated to all applications.
- Manage spend to budget
- Create budgets and budget alerts for each group to ensure you stay in budget throughout the year.
- Key alerts
- Forecasted month end cost exceeds alert threshold
- MTD cost is over alert threshold
- Forecasted year end cost exceeds alert threshold
- YTD cost is over alert threshold
- Sign up to receive monthly cost and usage reports for integration into your internal cost accounting system.
- Cost by application, environment, business unit etc.
Even though AWS’ “parts bin” approach to pricing is complicated, following these guidelines will help ensure accurate cost accounting of your cloud spend.
–Timothy Hill, Senior Product Manager, 2nd Watch
