Notes from the team building C3X. Cost estimation for Terraform, the economics of cloud infrastructure, and how to ship FinOps tooling without a SaaS gate.
The AWS free tier covers limited usage for twelve months (and some services always), and when it expires, resources that were free start billing at standard rates. Here is what changes, what to watch, and how to avoid a surprise bill.
Reserved instances save money on steady workloads but lock you into a commitment, so buying at the wrong time or for the wrong capacity wastes the discount. Here is how to decide when a workload is ready to reserve.
A high AWS bill almost always traces to a few dominant drivers, over-provisioned compute, data transfer, idle resources, and unmanaged growth, rather than a mystery. Here is how to find what is driving your bill and bring it down.
us-east-1 (N. Virginia) is AWS's oldest, largest region, and it is consistently at or near the lowest prices and the first to get new services. Here is why it earned that status, and when the cheapest region is not the right choice.
Running Apache Airflow on AWS costs money whether you use managed MWAA or self-host on EC2 or EKS. The bill is driven by environment size, worker scaling, and the compute your tasks actually launch. Here is how to size and control it.
Azure Kubernetes Service is free for the control plane, but node pools (the VMs running your workloads) are where the cost lives. Right-sizing pools, using spot and autoscaling, and separating workload types cut the bill substantially. Here is how.
AWS API Gateway offers REST APIs and HTTP APIs at very different prices. HTTP APIs are substantially cheaper per million requests but drop some features. Here is the price gap, the feature tradeoffs, and how to choose.
AWS AppSync (managed GraphQL) and API Gateway (REST and HTTP) are priced differently and suit different workloads. AppSync bills per query and real-time message; API Gateway bills per request. Here is how they compare and when each is cheaper.
AWS security groups are free, there is no charge for creating or using them. But the network resources they protect, and adjacent features like WAF and Network Firewall, do cost money. Here is what is free and what is not.
Some cloud data transfers are free (inbound, same-AZ private traffic) and some are expensive (egress to the internet, cross-region, cross-AZ). Knowing which is which is essential to avoiding surprise bills. Here is the map.
Both Athena and Redshift Spectrum query data in S3 and both charge per terabyte scanned, but their cost models and surrounding infrastructure differ. Here is how they compare and how to make either one cheap.
Aurora Serverless v2 scales database capacity in fine-grained Aurora Capacity Units and bills per ACU-hour. It can save money on variable workloads but cost more on steady ones. Here is how the pricing works and when to use it.
When your AWS bill jumps unexpectedly, a systematic investigation, Cost Explorer, grouping by service and usage type, checking recent changes, finds the cause fast. Here is a step-by-step troubleshooting playbook.
AWS Local Zones place compute and storage closer to end users in metro areas, at prices higher than the parent region. Here is how Local Zones are priced, the data transfer implications, and when the latency is worth the premium.
The identical EC2 instance, S3 bucket, or RDS database costs different amounts depending on the AWS region. Understanding why, and which regions are cheapest, can meaningfully cut your bill. Here is the full explanation.
AWS Wavelength embeds compute and storage inside telecom 5G networks for ultra-low latency to mobile devices, at a premium over the parent region. Here is how Wavelength is priced, the transfer implications, and when it is worth it.
Azure Blob storage offers hot, cool, cold, and archive access tiers at very different prices, trading storage cost against access cost and retrieval latency. Matching data to the right tier can cut storage spend dramatically. Here is how.
The Azure free tier mixes a 12-month trial, a set of always-free services, and a starting credit. Knowing which is which keeps you from an unexpected bill when a trial window closes. Here is the full breakdown.
Azure Machine Learning cost is dominated by compute, GPUs for training, endpoints for inference, and idle clusters. Right-sizing, autoscaling to zero, spot compute, and batch endpoints are the biggest levers. Here is the playbook.
Azure prices the same service differently by region, sometimes by a wide margin. The cause is local infrastructure cost, demand, and taxes. Knowing the pattern helps you place workloads where they are cheapest without hurting latency.
Azure reservations give a large discount on VMs, databases, and other services in exchange for a one or three year commitment. Understanding scope, flexibility, and exchange rules is key to committing without overcommitting. Here is how they work.
For most AWS accounts, a short list dominates the bill: compute, data transfer, storage, databases, and managed services. Knowing which drivers matter lets you focus optimization where it pays. Here is the breakdown.
BigQuery cost splits into query processing (bytes scanned or slots) and storage. Partitioning, clustering, avoiding SELECT star, and choosing on-demand versus capacity pricing are the biggest levers. Here is the detailed playbook.
FinOps succeeds as a culture, not a tool. Shared ownership, cost visibility for engineers, blameless accountability, and celebrating savings turn cost into a normal engineering concern. Here is how to build that culture.