As part of planning to migrate Pi-Hole from my Pi-400 to a highly available setup on Proxmox, I conducted DNS performance benchmarks to ensure the new LXC containers would provide equivalent performance to the existing Pi-400 installation. This testing is part of my broader 2026 homelab upgrade project.
This post documents the benchmarking methodology and results from testing various container configurations against the current Pi-400 setup to validate that performance would remain consistent during the migration.
Testing Methodology
All benchmarks were conducted using Steve Gibson’s DNS Benchmark tool with the “100x” setting and a fixed 100ms graph scale for consistent measurement across all runs. The testing followed Pi-Hole’s official benchmarking guidance from their documentation.
Test Environment
- Benchmark Tool: Steve Gibson’s DNS Benchmark
- Client: Laptop running benchmark queries
- Network: Mixed testing over WiFi and Ethernet connections
- Targets: Pi-400 baseline vs Proxmox LXC containers
- Graph Scale: Fixed at 100ms for consistent measurement across all runs
Benchmark Results
Performance Summary Table
| Run | Configuration | Network | DNS Server | Average Response Time (ms) |
|---|---|---|---|---|
| 1 | Baseline LXC | WiFi | Pi-400 | 40.84 |
| Router | 40.92 | |||
| Container 2 | 58.85 | |||
| Container 1 | 59.89 | |||
| 2 | Logging Disabled (Partial) | WiFi | Pi-400 | 40.31 |
| Router | 43.08 | |||
| Container 1 | 43.19 | |||
| Container 2 | 43.42 | |||
| 3 | Full Logging + Cache Tuning | WiFi | Router | 41.64 |
| Pi-400 | 42.22 | |||
| Container 2 | 49.96 | |||
| Container 1 | 61.70 | |||
| 4 | Resource Scaling (2 cores, 1GB) | WiFi | Pi-400 | 41.04 |
| Container 1 | 41.59 | |||
| Router | 42.28 | |||
| Container 2 | 55.56 | |||
| 5 | Ethernet Connection | Ethernet | Container 1 | 36.44 |
| Pi-400 | 38.84 | |||
| Router | 39.07 | |||
| Container 2 | 49.06 |
Detailed Results by Run
Run 1: Baseline Configuration
- Setup: Default LXC container settings, WiFi connection
- Best performer: Pi-400 at 40.84ms
- Key observation: Container performance was consistent with only minor variation between instances
Run 2: Logging Disabled (Partial)
- Setup: Query logging disabled on containers only (Pi-400 unchanged)
- Best performer: Pi-400 at 40.31ms
- Key finding: Significant container performance improvement - Container 1 improved 28% (59.89ms → 43.19ms), Container 2 improved 26% (58.85ms → 43.42ms)
Run 3: Full Logging Optimization + Cache Tuning
- Setup: Complete logging disable + 25,000 cache size
- Best performer: Router DNS at 41.64ms
- Key finding: Large cache size (25,000) did not improve performance, confirming Pi-Hole’s guidance
Run 4: Resource Scaling Test
- Setup: Increased to 2 cores, 1GB RAM, logging disabled
- Best performer: Pi-400 at 41.04ms
- Performance impact: Minimal improvement from resource scaling
Run 5: Network Impact Analysis
- Setup: Ethernet connection to Proxmox switch
- Best performer: Container 1 at 36.44ms
- Network impact: 10.7% performance improvement switching from WiFi to Ethernet
Migration Validation Results
Performance Equivalency Assessment
The primary goal was to confirm that Proxmox LXC containers would deliver DNS response times roughly equivalent to the existing Pi-400 setup. The testing aimed to validate that the containerized deployment would provide comparable performance characteristics.
Network Connectivity Impact
Switching from WiFi to Ethernet (Run 4 vs Run 5) showed a 10.7% improvement in response times, demonstrating that network connectivity does impact DNS performance but was not the most significant factor observed during testing.
Resource Requirements Validation
Testing with different resource allocations (Run 1 vs Run 4) showed that doubling CPU cores and RAM provided minimal performance gains. This confirmed that modest resource allocation would be sufficient for the migration.
Configuration Optimization Testing
The cache size experiment in Run 3 validated Pi-Hole’s documentation warnings about excessive cache sizes. Default settings proved to be appropriately tuned for typical home network DNS loads.
Container Performance Consistency
Testing revealed that container performance could vary significantly depending on configuration changes, with Container 1 showing more dramatic swings than Container 2 across different test scenarios.
Logging Overhead Assessment
Disabling query logging on the containers showed significant performance improvements - Container 1 improved 28% and Container 2 improved 26% between Run 1 and Run 2. This demonstrates that logging overhead can be substantial for containerized Pi-Hole deployments, unlike the Pi-400 which maintained consistent performance.
Migration Validation Conclusion
The benchmark results successfully validated that Proxmox LXC containers can deliver DNS performance equivalent to the existing Pi-400 setup. With response times consistently in the 36-42ms range across different configurations, the containerized deployment meets the performance requirements for the planned migration.
These results provide confidence that migrating Pi-Hole to a highly available Proxmox-based setup will maintain DNS performance while adding the reliability benefits of redundant containers across the cluster.
Costs
I purchased Steve Gibson’s DNS Benchmark tool for $9.95 to conduct this testing.