Memory Management Techniques

• Week 8, Part II focuses on memory management.

Memory Management

• Involves swapping blocks of data from secondary storage.
• Allows one part for the operating system such as the resident monitor or kernel.
• Allows one part for the program currently being executed.
• Subdivides the "user" part of memory in a multiprogramming system to accommodate multiple processes.
• Aims to maximize CPU efficiency by timing the swapping.
• Involves managing the need for more memory as applications become more demanding.
• Memory I/O is slow compared to a CPU.

Memory Management Terms

• Frame: A fixed-length block of main memory.
• Page: A fixed-length block of data in secondary memory that may be copied temporarily into a frame of main memory.
• Segment: A variable-length block of data in secondary memory that may be divided into pages or copied entirely into main memory.

Memory Management Requirements Include:

• Relocation capabilities
• Protection measures
• Sharing abilities
• Logical organization
• Physical organization

Relocation

• Active processes must be able to be swapped in and out of main memory in order to maximize processor utilization.
• Programmers do not typically know in advance which other programs reside in main memory during execution.
• Processes may need to relocate to a different area of memory.
• Specifying that a process must be placed in the same memory region when swapped back in is limiting.

Protection

• A program's location in main memory is unpredictable.
• Processes need permission to reference memory locations for reading or writing.
• Memory references made by a process must be checked at run time.
• Mechanisms supporting relocation also support protection.

Sharing

• Allows processes to access the same program copy instead of having separate copies.
• Mechanisms used to support relocation assist sharing capabilities.
• Controlled access to shared memory areas is required without protection compromise.

Logical Organization

• Memory is typically organized as linear.
• Programs are written in modules that can be independently written and compiled.
• Modules can be assigned protection levels such as read-only or execute-only.
• Module-level sharing corresponds to the user's problem view.
• Segmentation is the tool that best satisfies these logical requirements.

Physical Organization

• Programmers shouldn't manage memory
• Memory available for a program plus its data may be insufficient.
• Programmers don't know how much space will be available
• Overlaying assigns various modules to the same memory region but is time-consuming for the program.

Memory Partitioning

• Memory management brings processes into main memory for processor execution.
• It uses virtual memory.
• It is Based on segmentation and paging.
• Partitioning is used in obsolete operating systems like IBM's mainframe OS, OS/MFT (Multiprogramming with a Fixed Number of Tasks).
• It does not use virtual memory.

Memory Management Techniques

Fixed Partitioning

• Main memory is divided into static partitions at system startup; a process loads into a partition of equal or greater size.
• Simple to implement and puts little overhead on the operating system.
• The amount of active processes is fixed and is an inefficient use of memory due to internal fragmentation.

Dynamic Partitioning

• Partitions are created dynamically so that each process is loaded into a partition exactly the same size.
• It is a more efficient use of main memory with no internal fragmentation.
• Inefficient processor use due to compaction needed to counter external fragmentation.

Simple Paging

• Main memory is divided into equal-sized frames, with each process divided into equal-sized pages of the same length as frames.
• Processes are loaded by loading all pages into available, not always contiguous, frames.
• A small amount of internal fragmentation will occur, and there is no external fragmentation.

Simple Segmentation

• Each process is divided into a number of segments, loaded into dynamic partitions with no need for contiguity.
• Improved memory utilization with reduced overhead compared to dynamic partitioning, no internal fragmentation.
• External fragmentation occurs.

Virtual Memory Paging

• As with simple paging, it is not necessary to load all of the pages of a process. Nonresident pages that are needed are brought in later automatically.
• Higher degree of multiprogramming and a larger virtual address space with no external fragmentation.
• Overhead of complex memory management is required.

Virtual Memory Segmentation

• As with Segmentation, it is not necessary to load all of the segments of a process. Nonresident segments that are needed are brought in later automatically.
• Higher degree of multiprogramming, a very large virtual address space, plus it supports protection and sharing and has no internal fragmentation.
• Overhead of complex memory management is required.

Equal Sized, Fixed Partitioning Disadvantages

• A program may be too big to fit in a partition, needing overlays.
• Main memory utilization is inefficient as any program occupies an entire partition, regardless of size.
• Internal fragmentation results in wasted space when loaded data is smaller than the partition.

Unequal Sized, Fixed Partitioning Disadvantages

• The number of partitions specified at system generation time limits the number of active processes.
• Small jobs will not utilize partition space efficiently.

Dynamic Partitioning Issues

• External fragmentation occurs where memory becomes more and more fragmented, declining memory utilization
• Compaction shifts processes to be contiguous, putting free memory in one block, but is time-consuming and waists CPU time.