SOP-003: Configure Autoscaling

DOCUMENT CONTROL

Field	Value
SOP ID	SOP-003
Version	1.0
Status	Active

Purpose

Configure Cloud Run autoscaling to optimize cost and performance. Control how your service scales up and down based on traffic demands.

Key Concepts

Term	Description
Concurrency	Max simultaneous requests per instance
Min Instances	Always-on instances (0 = scale to zero)
Max Instances	Upper scaling limit
Cold Start	Time to start new instance from zero

Flowchart

                 ┌──────────────┐
                 │   Request    │
                 │   Arrives    │
                 └──────┬───────┘
                        │
         ┌──────────────┴──────────────┐
         │                             │
         ▼                             ▼
┌─────────────────┐         ┌─────────────────┐
│ Existing Inst.  │         │  No Available   │
│   Available?    │         │    Instance     │
└────────┬────────┘         └────────┬────────┘
         │ Yes                       │
         ▼                           ▼
┌─────────────────┐         ┌─────────────────┐
│ Route to Inst.  │         │  Cold Start     │
│ (if < concur.)  │         │  New Instance   │
└─────────────────┘         └────────┬────────┘
                                     │
                                     ▼
                            ┌─────────────────┐
                            │ Route Request   │
                            └─────────────────┘

Procedure

Step 1: Analyze Current Traffic Patterns

# View current metrics
gcloud run services describe my-service \
  --region us-central1 \
  --format 'yaml(status.traffic,spec.template.spec.containerConcurrency)'

Step 2: Configure Concurrency

CONCURRENCY GUIDELINES

• CPU-bound tasks: Lower concurrency (1-10)
• I/O-bound tasks: Higher concurrency (50-100)
• Default: 80 concurrent requests

# Set concurrency limit
gcloud run services update my-service \
  --region us-central1 \
  --concurrency 100

Step 3: Set Instance Limits

# Configure scaling bounds
gcloud run services update my-service \
  --region us-central1 \
  --min-instances 1 \
  --max-instances 10

COST IMPLICATIONS

Setting --min-instances > 0 means you'll always pay for at least that many instances, even with no traffic.

Step 4: Configure CPU Allocation

# Always-on CPU (recommended for latency-sensitive apps)
gcloud run services update my-service \
  --region us-central1 \
  --cpu-boost \
  --no-cpu-throttling

Mode	When CPU Allocated	Best For
Default	Only during request	Cost optimization
--no-cpu-throttling	Always on	Background tasks
--cpu-boost	Extra at startup	Faster cold starts

Step 5: Set Startup/Shutdown Timeouts

gcloud run services update my-service \
  --region us-central1 \
  --timeout 300 \
  --startup-cpu-boost

Scaling Strategies

Cost-Optimized (Scale to Zero)

gcloud run services update my-service \
  --region us-central1 \
  --min-instances 0 \
  --max-instances 5 \
  --concurrency 80

Low-Latency (Always Warm)

gcloud run services update my-service \
  --region us-central1 \
  --min-instances 2 \
  --max-instances 100 \
  --concurrency 50 \
  --cpu-boost

High-Throughput (Max Concurrency)

gcloud run services update my-service \
  --region us-central1 \
  --min-instances 1 \
  --max-instances 1000 \
  --concurrency 1000 \
  --memory 4Gi \
  --cpu 4

Verification Checklist

• [ ] Concurrency matches application capability
• [ ] Min instances set based on latency requirements
• [ ] Max instances within budget constraints
• [ ] CPU allocation mode appropriate for workload
• [ ] Monitored in Cloud Monitoring dashboard

Monitoring Autoscaling

# View instance count over time
gcloud monitoring dashboards list

# Check current instance count
gcloud run services describe my-service \
  --region us-central1 \
  --format 'value(status.conditions)'

Cost Calculator

Instances	Hours/Month	vCPU Cost	Memory Cost (1GB)
0 (scale to zero)	0	$0	$0
1	730	~$47	~$5
2	1460	~$94	~$10
5	3650	~$236	~$26

Prices approximate for us-central1. Use GCP Calculator for exact pricing.