Cloud Cost Engineering and Open Source FinOps
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.
Transit Gateway cost explained: attachments, data, and vs VPC peering
Transit Gateway is $0.05/hour per attachment (~$36.50/month each) + $0.02/GB. Here's why it costs more than peering, when the hub is worth it (~4-6 VPCs), and how to cut it.
AWS Backup cost explained: retention is the whole bill
AWS Backup bills mostly for backup storage (GB-month) plus restore and cross-region copy. The bill grows from retention creep. Here's how the meters work and how to keep backups from piling up forever.
Azure App Service Plan tiers cost: you pay for the plan, not the app
App Service Plans bill per instance-hour by tier (B1 ~$13, S1 ~$70, P1v3 ~$120+/mo) times instances. Apps share a plan for free. Here's how to pick the tier and why consolidating apps is the big saving.
Azure Front Door vs CDN cost: caching or full app delivery?
Azure CDN is pure caching billed per GB egress with no base fee; Front Door adds load balancing and WAF for a monthly base plus requests. Here's which to use and why cache hit ratio drives the bill.
GCP persistent disk types cost: pd-standard vs balanced vs ssd
Persistent disks bill per provisioned GB-month by type: pd-standard ~$0.04, pd-balanced ~$0.10, pd-ssd ~$0.17. Here's which to use, why you pay for provisioned size, and how to stop over-paying for SSD.
GCP Spot VMs: 60-91% off if you design for preemption
Spot VMs run 60-91% below on-demand for the same machine type, preemptible at any time. Here's what belongs on Spot, how it differs from preemptible, Spot GKE pools, and how to design for preemption.
Step Functions cost: Standard vs Express and the 100x difference
Standard workflows bill $0.025 per 1,000 state transitions; Express bills per request + duration. For high-volume workflows the gap is 50-100x. Here's which to use and why high-frequency Standard workflows blow up.
Azure Functions cost: Consumption vs Premium plan
Consumption bills per execution and scales to zero; Premium bills ~$150/month per always-on instance for no cold starts and VNet. Here's the crossover and the min-instances trap.
Cloud SQL vs AlloyDB cost: when the faster database is also cheaper
AlloyDB costs more per vCPU than Cloud SQL but is much faster for analytical workloads, so it can need less compute for the same performance. Here's how each bills and which to pick.
Secrets Manager vs Parameter Store cost: stop paying $0.40 for static secrets
Secrets Manager is $0.40/secret/month + API calls; Parameter Store standard tier is free. Here's what the fee actually buys (rotation, cross-account) and the split that cuts a large secrets bill 90%.
Aurora Serverless v2 cost: what an ACU really costs you
Aurora Serverless v2 bills per ACU-hour (~$0.12) scaling in 0.5-ACU steps, with a ~$44/month floor unless you enable scale-to-zero. Here's when it beats provisioned Aurora and when it doesn't.
OpenSearch cost optimization: shrink the always-on cluster
OpenSearch bills data-node hours + EBS + master nodes 24/7, so a small cluster is $700+/month. Here's how to cut it with UltraWarm tiering, Graviton nodes, gp3, lifecycle management, and Reserved Instances.
Azure Cosmos DB RU cost optimization: stop over-provisioning throughput
Cosmos DB bills by provisioned RU/s (~$0.008 per 100 RU/s-hour) times region count. Over-provisioned throughput and needless multi-region writes are the big overspends. Here's how to right-size and which throughput model to pick.
BigQuery cost: on-demand vs slots, and how to stop scanning terabytes
BigQuery on-demand bills ~$6.25/TB scanned; capacity bills for slots. Here's which model is cheaper for your query volume, and why partitioning and never SELECT * is the biggest saving on either.
AWS Graviton migration: where the ~20% savings actually come from
Graviton (ARM) instances list ~20% cheaper than x86 with up to 40% better price-performance, across EC2, Lambda, RDS, and more. Here's the discount, where it applies, and how small the migration usually is.
AWS WAF cost explained: the three meters that drive your bill
AWS WAF charges $5/month per Web ACL + $1/month per rule + $0.60 per million requests, plus paid add-ons like Bot Control. Here's how the meters stack and how to keep them down.
Azure storage account tiers cost: Hot vs Cool vs Cold vs Archive
Azure Blob tiers trade storage price for access price, and each cold tier adds a retention penalty. Here's the full picture, storage rates, access costs, minimum retention, and redundancy, and how to pick the right tier.
GCP network egress cost explained: ingress is free, everything out costs
GCP charges for egress to the internet, between regions, and even between zones, but not ingress or same-zone traffic. Here's the egress map, the three patterns that drive the bill, and how to cut it.
Fargate vs EC2 cost: the crossover is utilization, not the rate
Fargate costs 15-50% more per vCPU and GB than EC2, but you pay for zero idle capacity. Here's where the crossover actually is, why dense fleets favor EC2, and when Fargate's premium is the cheapest line on the bill.
ALB vs NLB cost: why they're nearly identical and what actually drives the bill
ALB and NLB both cost ~$16.43/month to exist plus LCU/NLCU usage at ~$0.006-0.008 per unit-hour. The price difference is small, the real lever is the LCU dimension your traffic stresses and how many balancers you run.
AKS cost optimization: the control plane is free, the nodes are not
The AKS control plane is free (or ~$73/month for the SLA tier), but worker node VMs are the real bill. Here's how to cut AKS cost with the autoscaler, Spot node pools, right-sizing, reservations, and scale-to-zero.
Cloud Run vs GKE cost: when serverless beats a node pool
Cloud Run bills per request-second and scales to zero; GKE bills for nodes 24/7. The cheaper option comes down to utilization. Here's the crossover, the min-instances trap, and where Autopilot fits.
ElastiCache vs MemoryDB cost: when durability is worth paying for
MemoryDB costs more than ElastiCache for the same node type and adds a per-GB data-written charge ElastiCache doesn't have. Here's the price difference, the durability you're buying, and when each is the right call.
RDS Proxy cost: is the always-on charge worth it?
RDS Proxy bills per vCPU of the database it fronts, continuously, whether or not connections flow. Here's how the charge works, when it pays for itself, and when it's overhead you can skip.