Skip to content

Google Cloud Cost Optimization

Overview

This guide covers Google Cloud cost management tools, strategies, and best practices. Cost optimization is relevant to Professional Cloud Architect and FinOps certification exams.


GCP Cost Management Tools

Tool Purpose Access
Billing Reports Visualize spending by project/service Cloud Console
Billing Export to BigQuery Detailed billing analysis BigQuery
Budgets and Alerts Spending thresholds and notifications Cloud Console
Recommender Right-sizing and optimization suggestions Cloud Console / API
Pricing Calculator Estimate costs before deployment Web tool
Cost Table Detailed cost breakdown Cloud Console
Active Assist Unified recommendations hub Cloud Console
FinOps Hub Cost governance and optimization Cloud Console

Compute Cost Optimization

Compute Engine Pricing Models

Model Discount Commitment Best For
On-Demand 0% None Short-term, unpredictable
Sustained Use Discounts Up to 30% Automatic (usage-based) Consistent monthly usage
Committed Use Discounts (CUDs) Up to 57% (compute), 70% (memory) 1 or 3 year Predictable workloads
Spot VMs Up to 91% None (preemptible) Fault-tolerant, batch
Flex CUDs Up to 46% 1 year (flexible) Variable instance types

Right-Sizing

Strategy Tool Action
VM right-sizing Recommender Resize based on utilization
Idle VM detection Recommender Delete or stop idle VMs
Schedule VMs Instance Scheduler Stop dev/test after hours
Custom machine types Compute Engine Match exact CPU/memory needs
Tau VMs (T2D/T2A) Compute Engine Best price-performance for scale-out
Sole-tenant nodes Compute Engine Licensing compliance at scale

Storage Cost Optimization

Cloud Storage Classes

Class Use Case Cost (per GB/month, US multi-region)
Standard Frequently accessed ~$0.026
Nearline Infrequent (30+ days) ~$0.01
Coldline Rare access (90+ days) ~$0.004
Archive Long-term (365+ days) ~$0.0012

Storage Strategies

Strategy Action Savings
Object lifecycle rules Auto-transition by age 50-95%
Autoclass Automatic class management Optimized per-object
Regional vs multi-regional Choose based on access pattern 50% (regional vs multi)
Delete unused snapshots Clean up Persistent Disk snapshots Variable
Persistent Disk type pd-standard vs pd-balanced vs pd-ssd 70% (standard vs ssd)

Database Cost Optimization

Strategy Service Action
CUDs Cloud SQL, Spanner, Bigtable 1 or 3 year commitments
Right-size instances Cloud SQL Match CPU/memory to workload
Read replicas Cloud SQL Offload reads
Spanner auto-scaling Spanner Scale processing units based on load
BigQuery flat-rate BigQuery Editions (Standard, Enterprise, Enterprise Plus)
BigQuery storage optimization BigQuery Partitioning, clustering, long-term storage
Firestore usage Firestore Optimize read/write patterns

BigQuery Cost Optimization

Strategy Description Impact
Partition tables Partition by date/integer Reduce data scanned
Cluster tables Cluster by frequently filtered columns Reduce data scanned
Use LIMIT with preview Avoid scanning full tables Reduce query cost
Materialized views Cache expensive query results Reduce repeated query cost
BI Engine In-memory analysis acceleration Fixed cost, fast queries
Editions pricing Autoscale or baseline slots Predictable cost
Long-term storage Auto-discount after 90 days 50% storage discount

Networking Cost Optimization

Strategy Description Savings
Standard Network Tier Lower cost egress (regional) 25-50% vs Premium
Cloud CDN Cache at edge, reduce origin Reduced egress
Private Google Access Access GCP APIs without external IP Avoid NAT costs
Cloud Interconnect Lower egress vs internet 70-80% egress savings
Same-zone traffic Co-locate resources Free intra-zone traffic
Cloud NAT optimization Right-size NAT bandwidth Reduced NAT charges

GCP-Specific Savings

Sustained Use Discounts (Automatic)

Usage (% of month) Discount
0-25% 0%
25-50% 20%
50-75% 40%
75-100% 60%
Full month effective ~30% average

Note: SUDs apply automatically to Compute Engine and Cloud SQL. No commitment required.

Committed Use Discounts

Type Duration Discount
Compute CUD 1 year Up to 37%
Compute CUD 3 year Up to 57%
Memory CUD 1 year Up to 50%
Memory CUD 3 year Up to 70%
Flex CUD 1 year Up to 46%

Organization-Level Optimization

Strategy Description
Billing accounts Organize by department/team
Labels Tag resources for cost allocation
Billing export Export to BigQuery for analysis
Budgets Set per-project budgets with alerts
Quotas Prevent cost overruns
Recommender Organization-wide recommendations
Custom dashboards Looker Studio with billing data

Cost Optimization Checklist

  • Enable billing export to BigQuery
  • Set up budgets and alerts per project
  • Review Recommender suggestions weekly
  • Implement Cloud Storage lifecycle rules
  • Analyze CUD and Flex CUD recommendations
  • Right-size Compute Engine instances quarterly
  • Use custom machine types where applicable
  • Schedule non-production VMs to stop after hours
  • Optimize BigQuery with partitioning and clustering
  • Delete unused persistent disks and snapshots
  • Evaluate Standard vs Premium Network Tier
  • Use Spot VMs for fault-tolerant workloads

  • GCP Billing: https://cloud.google.com/billing/docs
  • GCP Pricing Calculator: https://cloud.google.com/products/calculator
  • Committed Use Discounts: https://cloud.google.com/compute/docs/instances/signing-up-committed-use-discounts
  • GCP Recommender: https://cloud.google.com/recommender/docs
  • GCP Cost Optimization: https://cloud.google.com/architecture/framework/cost-optimization