Optimizing Linux Server Settings for Enhanced ClickHouse Performance: A Guide for High-Volume Data Ingestion

Shiv IyerShiv Iyer
3 min read

Optimizing a Linux server for ClickHouse, especially to handle high-velocity, high-volume data ingestion, involves several layers of system tuning. These enhancements are designed to maximize the performance of ClickHouse by leveraging the full potential of the underlying Linux system. Here are practical tips and tricks for tuning a Linux server specifically for ClickHouse performance:

1. Increase File Descriptors

ClickHouse can open a lot of files simultaneously, especially in high-load environments. Increase the number of available file descriptors to prevent "Too many open files" errors.

  • Edit Limits Configuration:

      # Edit /etc/security/limits.conf
      * soft nofile 262144
      * hard nofile 262144
    
  • Apply Changes:

      # For the changes to take effect without rebooting
      ulimit -n 262144
    

2. Optimize Network Settings

To improve the handling of high volumes of incoming connections and data, optimize the TCP stack:

  • Increase the Backlog and Buffers:

      # Edit /etc/sysctl.conf
      net.core.somaxconn = 4096
      net.core.netdev_max_backlog = 10000
      net.ipv4.tcp_max_syn_backlog = 4096
      net.core.rmem_max = 16777216
      net.core.wmem_max = 16777216
    
  • Apply Network Changes:

      sysctl -p
    

3. Adjust I/O Scheduling

The default I/O scheduler might not be optimal for database workloads. Changing the scheduler to deadline or noop can improve performance for SSDs:

  • Change Scheduler for SSDs:

      echo 'deadline' > /sys/block/sda/queue/scheduler
    

4. Optimize File System

Using the XFS or EXT4 file systems can enhance performance. XFS is particularly recommended for its scalability and performance with large files:

  • Mount Options: When mounting file systems, use options that reduce latency and improve throughput:

      # For example, mounting an XFS file system
      mount -o noatime,nodiratime /dev/sda /var/lib/clickhouse
    

5. Control Swappiness

Swappiness controls the degree to which the system favors swap over RAM. A lower value is preferred for database systems to force the Linux kernel to use RAM more aggressively.

  • Reduce Swappiness:

      # Set swappiness to a lower value
      sysctl vm.swappiness=10
    

6. Tune CPU Frequency Scaling

Ensure that CPU frequency scaling is set to performance mode to prevent fluctuations in CPU clock speed, which can impact latency:

  • Set CPU to Performance Mode:

      # Apply to all CPUs
      for CPU in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor
      do
          echo performance > $CPU
      done
    

7. Disable Transparent Huge Pages (THP)

THP can cause performance degradation with databases due to how memory is managed. It's often better to disable it:

  • Disable THP:

      echo never > /sys/kernel/mm/transparent_hugepage/enabled
      echo never > /sys/kernel/mm/transparent_hugepage/defrag
    

8. Regular System Monitoring and Maintenance

Keep an eye on system metrics such as CPU usage, I/O wait, memory usage, and network throughput. Regularly updating the Linux kernel and system packages can also help maintain optimal performance.

By implementing these optimizations, you can significantly enhance the performance of ClickHouse on a Linux server, particularly in scenarios involving high data ingestion rates and volumes. Regularly review and adjust these settings based on the specific workloads and system behavior over time.

0
Subscribe to my newsletter

Read articles from Shiv Iyer directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Shiv Iyer
Shiv Iyer

Over two decades of experience as a Database Architect and Database Engineer with core expertize in Database Systems Architecture/Internals, Performance Engineering, Scalability, Distributed Database Systems, SQL Tuning, Index Optimization, Cloud Database Infrastructure Optimization, Disk I/O Optimization, Data Migration and Database Security. I am the founder CEO of MinervaDB Inc. and ChistaDATA Inc.