Memory Swapping

The dedicated secondary storage area on a disk (HDD or SSD) that holds memory pages evicted from RAM. It acts as an overflow area for physical memory.

• Swap File: A special file within a filesystem configured for swapping.
• Swap Partition: A dedicated disk partition used exclusively for swap, often offering slightly better performance.
• The operating system's memory manager handles the mapping between RAM frames and swap space locations.

The core data structure that tracks the location of each virtual memory page. For each page, the page table entry contains a present bit.

• Present Bit = 1: The page is resident in physical RAM.
• Present Bit = 0: The page is not in RAM; it resides in swap space. The entry then stores the disk address within the swap area.
• When the CPU accesses a page marked "not present," it triggers a page fault, prompting the OS to fetch it from swap.

A kernel subsystem that responds to page faults—interrupts generated by the Memory Management Unit (MMU) when a process accesses a page not currently in RAM.

Its primary swapping-related functions are:

• Swap-In (Page In): Locate the required page in swap space, find a free RAM frame (or evict one), load the page, and update the page table.
• Swap-Out (Page Out): Select a victim page in RAM, write it to swap space if modified (dirty page), mark its page table entry as "not present," and free the RAM frame. This handler is critical for making swapping transparent to running processes.

The policy that decides which page in RAM to evict (swap out) when a free frame is needed. The goal is to minimize future page faults.

Common algorithms include:

• Least Recently Used (LRU): Evicts the page that hasn't been accessed for the longest time.
• Clock (Second Chance): An efficient approximation of LRU using a reference bit.
• First-In, First-Out (FIFO): Evicts the oldest page. The choice of algorithm directly impacts system performance under memory pressure, as frequent swapping (thrashing) can cripple throughput.

A flag in the page table entry and hardware that indicates whether a page in RAM has been written to since it was last loaded from disk or swap.

This bit is crucial for swap efficiency:

• Dirty Page (Bit=1): Must be written back to swap space before its frame can be reused, as it contains new data not on disk.
• Clean Page (Bit=0): Can simply be discarded (overwritten) if a copy already exists in swap or the original file (e.g., program code). Tracking dirty pages prevents unnecessary write operations to the slower swap device.

A background kernel process (daemon) that proactively manages swap activity to avoid latency spikes during critical application execution.

Its functions include:

• Proactive Swapping: Monitors free memory levels and preemptively swaps out pages when memory is low but before the system is critically starved.
• Page Cache Management: Often works with the system's page cache, reclaiming clean cache pages before resorting to swapping application memory.
• Cluster Writing: Groups multiple pages to be swapped out into contiguous blocks, optimizing disk I/O throughput. This daemon helps smooth out performance and prevent sudden, disruptive thrashing.

The organization of memory and storage subsystems into multiple levels with distinct trade-offs between speed, capacity, and cost per bit. This fundamental computer architecture principle enables systems to appear both fast and capacious.

• Levels: Registers → L1/L2/L3 Cache → Main Memory (RAM) → Solid-State/Flash Storage → Hard Disk Drives → Tape/Archival.
• Principle: Frequently accessed data resides in faster, smaller, more expensive memory (e.g., RAM), while less active data resides in slower, larger, cheaper storage (e.g., disk).
• Swapping's Role: Implements the movement of data between the RAM and disk levels of this hierarchy based on demand.

A dynamic, automated storage management technique that moves data between different classes (tiers) of memory or storage media based on observed access patterns and predefined policies. It is a generalization of the swapping concept.

• Key Difference from Swapping: Tiering often works at a sub-page or object granularity and can involve multiple storage tiers (e.g., fast NVMe vs. slow HDD, or different RAM technologies).
• Policy-Driven: Uses algorithms to promote hot data (frequently accessed) to faster tiers and demote cold data to slower tiers.
• Use Case: Modern database systems and hypervisors use tiering to optimize performance for large, active datasets.

A memory management technique that provides an abstraction layer between the software's view of memory (virtual address space) and the actual physical RAM. It is the foundational system within which swapping operates.

• Core Mechanism: Uses a Page Table, managed by the Memory Management Unit (MMU), to translate virtual addresses to physical addresses.
• Enables Swapping: When a needed page is not in RAM (a page fault), the virtual memory system triggers the swap-in operation from disk.
• Benefits: Provides processes with the illusion of a large, contiguous, and private address space, simplifies programming, and enables memory isolation and protection.

A software mechanism in an operating system kernel that keeps recently accessed disk blocks (pages) in unused portions of RAM. It is a complementary optimization to swapping, working on the file system layer.

• Purpose: Dramatically speeds up repeated reads from and writes to slow disk storage by serving data from fast RAM.
• Interaction with Swapping: The page cache consumes RAM. When the system needs more RAM for applications, it can reclaim pages from the page cache (which is cheap) before resorting to swapping out application memory (which is expensive).
• Linux Example: The free command shows memory used for page cache under the "buff/cache" column.

The set of memory pages that a process actively needs within a given time interval to operate efficiently without excessive page faults. Managing the working set is critical to minimizing swap thrashing.

• Thrashing: Occurs when the system's total working set size exceeds available physical RAM, causing continuous swapping as pages are constantly evicted and recalled.
• Principle of Locality: The working set exists because of temporal locality (recently used pages will likely be used again) and spatial locality (pages near recently used pages will likely be used).
• System Tuning: Monitoring working set sizes helps in right-sizing VM RAM allocation and configuring swap space appropriately.

The two primary forms of swap space (the dedicated disk area used for memory swapping). They represent different configurations for the same underlying function.

• Swap Partition: A dedicated, contiguous disk partition formatted for use as swap space. It is generally slightly faster and more reliable, as it has no filesystem overhead.
• Swap File: A regular file within the root filesystem designated as swap space. It is more flexible—easily resized, added, or removed without repartitioning the disk.
• Modern Usage: Most modern Linux distributions use a swap file by default (e.g., /swapfile) for simplicity, with performance differences being minimal on SSDs.