Cloud Infrastructure & AWS Monthly Cost Estimate Markdown Table Template

When to Use This Table

Formatting This Table in Markdown

A cloud infrastructure & aws monthly cost estimate table puts a few things in cells that Markdown treats as syntax. These are the ones to watch so it renders correctly on GitHub, GitLab, Obsidian and docs sites.

• Two dollar amounts in one cell can trigger math

  The Unit Billing Rate cells hold two prices, such as $0.04048 / vCPU-hr + $0.004445 / GB-hr. On GitHub $…$ is math, so text between two dollar signs may render as a formula; escape each sign as \$, or put USD in the header and keep cells numeric.

• Move the unit out of the total cell

  A total like **$868.47 / mo** does not right-align with the digits above it. Put USD / mo in the header, keep every cost to two decimals, and let bold be the only difference in the total row; see column alignment.

• GFM has no total row: bold label plus placeholders

  There is no footer row, formula or colspan, so the total is an ordinary last row: a bold label in the first cell, — in unused cells, and a bold amount. Recompute it by hand or in a spreadsheet whenever a line changes, because nothing in Markdown sums a column.

• Write multiplication as × or in a code span

  In capacity cells like 3 tasks × 730 h, the × is plain Unicode. Typing 3*730*2 instead italicizes the 730, because a pair of asterisks is emphasis; use × or a code span for formulas.

What Goes in Each Column

How to Fill It In

1. 1Write the assumptions down before any numbers: region, currency, pricing model, hours per month, and the date. Put them in a short block above the table (see the first variation) so nobody has to guess later.
2. 2Build the inventory from infrastructure code, not from the architecture diagram. A terraform plan or CDK synth output lists every billable resource, including the NAT gateways, load balancers, and log groups that diagrams leave out. For an existing system, use Cost Explorer (or your provider’s billing export) grouped by service and usage type for the last full month.
3. 3Price each line in the provider’s own calculator, save the estimate, and paste its shareable link under the table. The calculator handles tiers and regional rates that are easy to get wrong by hand.
4. 4Estimate usage-driven lines from real metrics: egress from load balancer or CDN logs, request counts from access logs, storage growth from the last few months of bucket metrics. If there is no history yet, write down the traffic assumption and the headroom you added, so both can be challenged.
5. 5After the first full billing month, add an Actual column and explain every line that moved by more than a threshold your team agrees on. Tag resources by project from day one, otherwise the comparison will not be possible.

Ready-to-Paste Variations

| Assumption | Value |
| :--- | :--- |
| Region | `us-east-1` |
| Pricing model | On-demand list prices; no Savings Plans or reservations |
| Hours per month | 730 |
| Rates checked | YYYY-MM-DD in the provider’s pricing calculator (link) |
| Currency | USD, excluding tax and support plan |
| Traffic forecast | 1.2 TB/mo egress, 90-day average from CDN logs |

| Service | Billing dimension | Unit | Monthly quantity |
| :--- | :--- | :--- | ---: |
| Fargate | vCPU | vCPU-hour | 4,380 (6 vCPU × 730 h) |
| Fargate | Memory | GB-hour | 8,760 (12 GB × 730 h) |
| NAT Gateway | Gateway hours | hour | 1,460 (2 gateways × 730 h) |
| NAT Gateway | Data processed | GB | 1,200 |
| Data transfer | Out to internet | GB | 1,200 |
| Aurora | Storage | GB-month | 200 |
| Aurora | I/O requests (Standard config only) | million requests | 150 |

| Category | Estimate | Actual | Variance | Driver |
| :--- | ---: | ---: | ---: | :--- |
| Compute | $216 | $231 | +$15 (+6.9%) | One extra task during the load test |
| Database | $380 | $380 | $0 (0.0%) | As planned |
| NAT and data transfer | $87 | $142 | +$55 (+63.2%) | Container image pulls routed through NAT |
| Storage | $59 | $61 | +$2 (+3.4%) | Log retention longer than assumed |
| **Total** | **$742** | **$814** | **+$72 (+9.7%)** | |

Common Mistakes

• Leaving out the NAT gateway hourly charge and internet egress.

  A NAT gateway bills per hour per gateway and per GB processed, and traffic leaving for the internet is billed again as data transfer out. Give each of these its own row, and consider VPC endpoints for traffic to S3 or ECR that would otherwise go through NAT.

• Multiplying hourly rates by 720 hours.

  A 30-day month is 720 hours, but the AWS calculator averages to 730 (8,760 hours a year divided by 12). Use 730 so your figures match the calculator and the average invoice.

• Presenting rates without a date, region, or pricing model.

  Add the assumptions block and a link to the saved calculator estimate. When someone reuses the table a year later, they will know which rates to refresh instead of trusting stale numbers.

• Costing only the primary database instance.

  Count every replica or standby, storage and I/O (on Aurora Standard), backup storage beyond the free retention allowance, and manual snapshots. Managed databases bill on several dimensions besides instance hours.

• Mixing list prices with committed-use discounts in one column.

  Show on-demand cost first and put discounts in a separate column or row. Reviewers need to see what happens if the commitment lapses or the workload moves.