Master Linux Cache Tuning: Boost Performance with Proven Tools & Techniques

1. Understanding Linux Cache

Cache is a high‑speed storage acting as a temporary “transit station” for frequently accessed data, allowing programs to read from memory instead of slower disk, greatly improving access speed.

In Linux, cache (page cache and inode cache) accelerates server response by bridging slow disk I/O and fast CPU.

2. Cache Performance Metrics

Key metrics include cache hit rate, cache size, and update frequency. High hit rate (>90%) indicates efficient caching; low hit rate (<30%) signals problems. Tools such as cachestat and cachetop report hit statistics.

Cache size comprises file‑system cache and page cache; too large can exhaust memory, too small leads to frequent disk I/O. Update frequency balances data freshness against overhead.

3. Cache Tuning Tools

sysctl

adjusts kernel parameters at runtime, e.g., sudo sysctl -w vm.swappiness=10 to reduce swapping. free -h and vmstat display memory and I/O usage. cachestat and cachetop (from the bcc package) show cache hit/miss data.

4. Practical Tuning Steps

4.1 Check current cache status

Use free -h and vmstat 1 to observe buff/cache, bi, bo values.

4.2 Adjust kernel parameters

Set vm.swappiness (0‑100) lower (10‑30) to keep data in memory; tune vm.dirty_ratio (5‑10) and vm.dirty_background_ratio (1‑5) to control dirty page flushing.

4.3 Clean caches

Echo 1, 2, or 3 to /proc/sys/vm/drop_caches to drop page cache, dentries/inodes, or all caches after running sync.

4.4 Optimize applications

Read files in batches instead of line‑by‑line, cache computed results, and set appropriate HTTP cache headers (e.g., expires 1w for CSS/JS, expires 1M for images).

5. Case Study

5.1 Problem

Website slowdown with high bi / bo and large buff/cache usage.

5.2 Diagnosis

Low cache hit rate revealed by cachestat; frequent temporary file creation caused cache churn; default vm.swappiness=60 increased swapping.

5.3 Solution

Refactor code to reduce temporary files, lower vm.swappiness to 10, and clear caches with echo 3 > /proc/sys/vm/drop_caches during low‑traffic periods, resulting in higher hit rates and reduced I/O.

# 逐行读取
with open('large_file.txt', 'r') as f:
    while True:
        line = f.readline()
        if not line:
            break
        # 处理每一行数据

# 批量读取
with open('large_file.txt', 'r') as f:
    lines = f.readlines()
    for line in lines:
        # 处理每一行数据

# 缓存字典
fib_cache = {}
def fib(n):
    if n in fib_cache:
        return fib_cache[n]
    if n <= 1:
        result = n
    else:
        result = fib(n - 1) + fib(n - 2)
    fib_cache[n] = result
    return result

# Nginx cache for CSS/JS
location ~ \.(css|js)$ {
    expires 1w;
    # other config
}

# Nginx cache for images
location ~ \.(jpg|jpeg|png|gif)$ {
    expires 1M;
    # other config
}