Why Linux Seems to Eat Your RAM—and How Buffers & Cache Actually Boost Performance

Free Command Overview

The free command reports memory usage in three lines. The second line shows total, used, free, shared, buffers, and cached memory (in megabytes when using the -m option). The third line recalculates memory usage by subtracting buffers and cache, giving a view that aligns with what a user perceives as truly available memory.

Buffers vs. Cache

A buffer holds data that has not yet been written to disk.

A cache holds data that has already been read from disk and is kept in memory for faster subsequent access.

Buffers improve write performance, while cache speeds up read operations. Both are counted as used memory by the kernel but can be reclaimed quickly when applications need RAM.

Memory Classification

From the user’s perspective, memory can be grouped into three categories:

Used : memory occupied by applications.

Free : memory not allocated to anything.

Buffers/Cache : memory used by the kernel for caching; it appears as used but can be reclaimed, so users often treat it as free.

Consequently, the free command’s third line (‑/+ buffers/cache) provides a more realistic picture of memory available to user processes.

Benefits of Buffers and Cache

Linux employs two major caching mechanisms—Buffer Cache and Page Cache—to accelerate disk I/O. Data read from slow block devices is stored in memory; subsequent accesses can be satisfied directly from RAM, bypassing the disk and dramatically improving performance. When memory pressure arises, the kernel quickly drops cache pages before resorting to swap, preserving responsiveness.

Experiment Demonstrating Cache Impact

The author performed a simple experiment on a lab file server:

Created a 1 GB test file.

Cleared the system cache ( sync; echo 3 > /proc/sys/vm/drop_caches).

Read the file once and measured the elapsed time.

Read the same file a second time without clearing the cache and measured the time again.

Results (illustrated by the images below) showed the first read took about 18 seconds, while the second read completed in roughly 0.3 seconds—a 60‑fold speedup, confirming the dramatic benefit of cached data.

Conclusion

Linux does not “eat” RAM; instead, it intelligently uses otherwise idle memory for buffers and cache, dramatically improving disk I/O performance. Users should interpret the free command’s third line to assess real free memory, and recognize that a high‑usage figure is often a sign of effective caching rather than a problem.

References

1. Discussion on “cp command double‑memory usage” – https://goo.gl/heEybX

2. Hacker News article “Linux ate my ram” – https://goo.gl/6DokDN

3. Additional experiments on cache impact – https://goo.gl/axxcbi

4. Blog post introducing the free command – https://goo.gl/W1yQVr