proxmox/docs/10-best-practices/PERFORMANCE_TUNING.md

Performance Tuning Guide

Last Updated: 2025-01-20
Document Version: 1.0
Status: Active Documentation

---

Overview

This guide provides performance tuning recommendations for Proxmox infrastructure, including host optimization, VM/container optimization, storage optimization, and network optimization.

---

Host Performance Tuning

CPU Optimization

Settings:

1. CPU Governor:

   # Set performance governor
   cpupower frequency-set -g performance
   
   # Make permanent
   echo 'GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_pstate=disable"' >> /etc/default/grub
   update-grub
   

2. CPU Affinity:

   • Pin critical VMs to specific CPU cores
   • Isolate CPU cores for host operations
   • Use CPU sets for resource allocation

Memory Optimization

Settings:

1. Transparent Huge Pages:

   # Check current setting
   cat /sys/kernel/mm/transparent_hugepage/enabled
   
   # Set to always (for performance)
   echo always > /sys/kernel/mm/transparent_hugepage/enabled
   

2. Swappiness:

   # Reduce swappiness (default is 60)
   echo 10 > /proc/sys/vm/swappiness
   
   # Make permanent
   echo 'vm.swappiness=10' >> /etc/sysctl.conf
   

3. Memory Overcommit:

   # Allow memory overcommit (for Proxmox)
   echo 1 > /proc/sys/vm/overcommit_memory
   

Storage Optimization

ZFS Tuning:

1. ARC Size:

   # Set ARC max size (adjust based on RAM)
   echo 'options zfs zfs_arc_max=4294967296' >> /etc/modprobe.d/zfs.conf
   

2. ZFS Recordsize:

   # Set recordsize for database workloads
   zfs set recordsize=16k <pool>/<dataset>
   

3. ZFS Compression:

   # Enable compression (lz4 recommended)
   zfs set compression=lz4 <pool>/<dataset>
   

LVM Tuning:

1. I/O Scheduler:

   # Set to deadline or noop for SSDs
   echo deadline > /sys/block/sda/queue/scheduler
   

---

VM/Container Performance

VM Optimization

CPU Settings:

1. CPU Type:

   • Use host CPU type for best performance
   • Or use specific CPU model matching host

2. CPU Cores:

   • Allocate appropriate number of cores
   • Avoid over-allocation
   • Consider CPU pinning for critical VMs

Memory Settings:

1. Memory Allocation:

   • Allocate sufficient memory
   • Use ballooning for dynamic allocation
   • Monitor memory usage

2. Memory Ballooning:

   # Enable ballooning in VM config
   balloon: 1024
   

Storage Settings:

1. Disk Cache:

   • Use writeback cache for better performance
   • Use none for critical data integrity

2. Disk I/O:

   • Set appropriate I/O limits
   • Use SSD storage for high I/O workloads
   • Consider separate storage pools

Container Optimization

Resource Limits:

1. CPU Limits:

   # Set CPU limit in container config
   lxc.cgroup.cpuset.cpus = "0-3"
   

2. Memory Limits:

   # Set memory limit
   lxc.cgroup.memory.limit_in_bytes = 2147483648
   

3. I/O Limits:

   # Set I/O limits
   lxc.cgroup.blkio.weight = 500
   

---

Network Performance

Network Optimization

Settings:

1. Network Interface:

   • Use virtio network drivers
   • Enable SR-IOV if available
   • Tune network buffer sizes

2. Bridge Optimization:

   # Increase bridge forward delay
   echo 0 > /sys/class/net/vmbr0/bridge/forward_delay
   

3. TCP Tuning:

   # Increase TCP buffer sizes
   echo 'net.core.rmem_max = 16777216' >> /etc/sysctl.conf
   echo 'net.core.wmem_max = 16777216' >> /etc/sysctl.conf
   sysctl -p
   

---

Monitoring Performance

Key Metrics

1. CPU:

   • CPU utilization
   • CPU wait time
   • CPU steal time

2. Memory:

   • Memory usage
   • Swap usage
   • Memory pressure

3. Storage:

   • I/O wait
   • Disk utilization
   • I/O throughput

4. Network:

   • Network throughput
   • Packet loss
   • Latency

Monitoring Tools

1. Proxmox Built-in:

   • Resource usage graphs
   • Performance metrics
   • Historical data

2. External Tools:

   • Prometheus + Grafana
   • Zabbix
   • Custom monitoring scripts

---

Performance Benchmarks

Baseline Measurements

Before Optimization:

• Document current performance
• Identify bottlenecks
• Set performance targets

After Optimization:

• Measure improvements
• Document results
• Adjust as needed

Benchmark Tools

1. CPU:

   # CPU benchmark
   sysbench cpu --cpu-max-prime=20000 run
   

2. Memory:

   # Memory benchmark
   sysbench memory --memory-total-size=10G run
   

3. Disk:

   # Disk I/O benchmark
   fio --name=test --ioengine=libaio --iodepth=16 --rw=read --bs=4k --size=1G
   

---

Troubleshooting Performance Issues

Common Issues

1. High CPU Usage:

   • Check for runaway processes
   • Review CPU allocation
   • Consider CPU pinning

2. High Memory Usage:

   • Check for memory leaks
   • Review memory allocation
   • Enable ballooning

3. High I/O Wait:

   • Check disk I/O
   • Review storage configuration
   • Consider SSD storage

4. Network Latency:

   • Check network configuration
   • Review network drivers
   • Tune network settings

---

Best Practices

1. Regular Monitoring:

   • Monitor performance metrics
   • Identify trends
   • Proactive optimization

2. Baseline Measurements:

   • Document baseline performance
   • Track changes over time
   • Set performance targets

3. Gradual Optimization:

   • Make one change at a time
   • Measure impact
   • Document results

4. Resource Planning:

   • Plan for growth
   • Monitor resource usage
   • Scale proactively

---

Related Documentation

• MONITORING_SUMMARY.md - Monitoring setup
• TROUBLESHOOTING_FAQ.md - Troubleshooting
• RECOMMENDATIONS_AND_SUGGESTIONS.md - Best practices

---

Last Updated: 2025-01-20
Review Cycle: Quarterly