Swap Space Management - Virtual Memory

Introduction

Swap space is an area on secondary storage (disk) that the operating system uses as an extension of physical RAM. When the system runs low on physical memory, the OS moves less-used pages from RAM to swap space — a process called "swapping out" or "paging out." When those pages are needed again, they are loaded back from swap — "swapping in" or "paging in."

Think of swap space like overflow parking for a busy mall. The main parking lot (RAM) has limited spots. When full, cars (pages) that have been parked the longest without being used are moved to a remote lot (swap). When their owners return, the cars are brought back to the main lot — which takes longer but prevents the mall from refusing new visitors entirely.

Why Swap Space Exists

Problem: Physical RAM is Limited and Expensive

A system with 8GB RAM running multiple applications might need 12GB total. Without swap, the system must either refuse to start new programs or kill existing ones. With swap, the OS moves inactive pages to disk, freeing RAM for active work.

Benefits of Swap

1. Support processes larger than RAM: A single process can use more virtual memory than physical RAM available
2. Handle memory spikes: Temporary peaks in memory usage do not cause crashes
3. Efficient RAM utilization: Rarely-used pages (e.g., startup initialization code) moved to swap, freeing RAM for active data
4. Enable hibernation: System state (entire RAM contents) written to swap for later restoration
5. Memory overcommit: Linux can promise more memory than physically available, relying on not all allocations being fully used simultaneously

Swap Location: Partition vs. File

Swap Partition

A dedicated disk partition used exclusively as swap space:

Advantages: Slightly faster (no filesystem overhead), guaranteed contiguous space. Disadvantages: Fixed size (must repartition to change), must be set up during installation.

Swap File

A regular file on an existing filesystem designated as swap space:

Advantages: Easy to resize (create new, delete old), no repartitioning needed, flexible. Disadvantages: Slight overhead from filesystem layer, possible fragmentation.

Modern Linux (4.0+) handles swap files nearly as efficiently as partitions.

Swap Allocation Strategies

How Pages are Selected for Swapping

The OS uses page replacement algorithms (LRU approximation via the clock algorithm) to choose which pages to swap out:

1. Anonymous pages: Process data (heap, stack) that has no file backing — must go to swap when evicted
2. File-backed pages (page cache): Can be discarded and re-read from the original file — no swap needed
3. Dirty file pages: Modified cached file data — must be written back to the file before eviction

Linux Swappiness

The swappiness parameter (0-100) controls the kernel's tendency to swap out anonymous pages vs. dropping file cache:

Swap Priority

When multiple swap devices exist, priority determines which is used first:

Performance Implications

The Swap Death Spiral

When the system uses swap heavily, performance degrades dramatically:

Thrashing: When the working set exceeds available RAM, the system constantly swaps pages in and out. CPU utilization drops (processes waiting for I/O), disk utilization hits 100%, and the system becomes essentially unusable.

How Much Swap to Allocate

For servers with abundant RAM, swap serves as an emergency buffer rather than regular overflow. Some administrators disable swap entirely for latency-sensitive services (databases, real-time systems) and rely on memory limits to prevent overcommit.

Swap in Different Operating Systems

Linux

• Supports swap partitions and swap files
• Multiple swap devices with priorities
• zswap: Compressed swap cache in RAM (pages compressed before going to disk)
• zram: RAM-based compressed block device used as swap (no disk I/O)

Windows

• Uses a page file (pagefile.sys) — essentially a swap file
• System-managed or fixed size
• Located on the system drive by default (can be moved to faster drive)

macOS

• Uses swap files in /private/var/vm/
• Compressed memory (since macOS Mavericks) — similar to Linux zswap
• Dynamic sizing based on pressure

Monitoring Swap Usage

Key Takeaways

• Swap extends virtual memory by using disk space as RAM overflow
• Pages selected for swapping are typically least-recently-used anonymous pages
• Swap files are nearly as fast as partitions in modern Linux and easier to manage
• Heavy swap usage indicates insufficient RAM — it is an emergency measure, not a performance feature
• Swappiness controls the balance between evicting anonymous pages vs. file cache
• zswap/zram can dramatically improve swap performance by compressing pages in RAM
• Monitor swap activity — consistent swapping means the system needs more physical memory