Benchmarking

Measure and analyze DNS performance using industry-standard tools.

Tools

dnsperf

Industry-standard DNS benchmarking:

# Install dnsperf
apt-get install dnsperf

# Create query file
cat > queries.txt <<'QUERIES'
example.com A
www.example.com A
mail.example.com MX
QUERIES

# Run benchmark
dnsperf -s $SERVICE_IP -d queries.txt -l 60 -Q 1000

resperf

Response rate testing:

# Test maximum QPS
resperf -s $SERVICE_IP -d queries.txt -m 10000

dig

Simple latency testing:

# Measure query time
dig @$SERVICE_IP example.com | grep "Query time"

# Multiple queries for average
for i in {1..100}; do
  dig @$SERVICE_IP example.com +stats | grep "Query time"
done | awk '{sum+=$4; count++} END {print "Average:", sum/count, "ms"}'

Benchmark Scenarios

Scenario 1: Baseline Performance

Single client, sequential queries:

dnsperf -s $SERVICE_IP -d queries.txt -l 60 -Q 100

Expected: < 10ms latency, > 90% success

Scenario 2: Load Test

Multiple clients, high QPS:

dnsperf -s $SERVICE_IP -d queries.txt -l 300 -Q 5000 -c 50

Expected: < 50ms latency under load

Scenario 3: Stress Test

Maximum capacity test:

resperf -s $SERVICE_IP -d queries.txt -m 50000

Expected: Find maximum QPS before degradation

Metrics to Collect

Response Time

• Minimum latency
• Average latency
• 95th percentile
• 99th percentile
• Maximum latency

Throughput

• Queries per second
• Successful responses
• Failed queries
• Timeout rate

Resource Usage

# During benchmark
kubectl top pods -n dns-system

# CPU and memory trends
kubectl top pods -n dns-system --use-protocol-buffers

Sample Benchmark Report

Benchmark: Load Test
Date: 2024-11-26
Duration: 300 seconds
Target QPS: 5000

Results:
- Queries sent: 1,500,000
- Queries completed: 1,498,500
- Success rate: 99.9%
- Average latency: 12.3ms
- 95th percentile: 24.1ms
- 99th percentile: 45.2ms
- Max latency: 89.5ms

Resource Usage:
- Average CPU: 1.2 cores
- Average Memory: 512MB
- Peak CPU: 1.8 cores
- Peak Memory: 768MB

Continuous Benchmarking

Automated Testing

apiVersion: batch/v1
kind: CronJob
metadata:
  name: dns-benchmark
spec:
  schedule: "0 2 * * *"  # Daily at 2 AM
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: dnsperf
            image: dnsperf:latest
            command:
            - /bin/sh
            - -c
            - dnsperf -s primary-dns -d /queries.txt -l 60 >> /results/benchmark.log

Trend Analysis

Track performance over time:

• Daily benchmarks
• Compare before/after changes
• Identify degradation early
• Capacity planning

Best Practices

1. Consistent tests - Same queries, duration
2. Isolated environment - Minimize external factors
3. Multiple runs - Average results
4. Document changes - Link to config changes
5. Realistic load - Match production patterns

Next Steps

• Performance - Performance overview
• Tuning - Optimization parameters