Virtual Memory And Its Architecture In Operating

Systems Essay, Research Paper

INTRODUCTION

Virtual memory refers to an imaginary set of locations, or addresses, where you can store data. It is imaginary in the sense that the memory area is not the same as the real physical memory composed of transistors or capacitors. The efficient use of memory is a problem that has remained important since the development of the stored_program computer. At a time when computer memory was still expensive and had a small capacitance the programmers were encouraged to develop programs, that used less than expected memory. This problem led to one of the most ingenious ideas. In 1961 a group of people at Manchester, England proposed a method for performing the overlay process automatically which, in turn, led to virtual memory.

The concept of virtual memory is to greatly increase the apparent amount of physical memory available to programs. The free space on the computer.s hard drive is treated as an extension of RAM (random access memory). When necessary, pages are swapped out of actual memory to a location on the hard drive. The swapped files can grow and shrink as thew memory demand fluctuates.

This concept is the key, which allows a user to run an operating system on a machine with 4, or 8 MB of installed RAM. On such systems, the operating system must swap portions of its self from the physical RAM to the hard drive.s swap files. This can also be done as it loads. In addition, the virtual memory also utilizes the processor.s ability to stop executing instructions when a page translation fails. In this paper the Unix, Apple, and the NT operating systems will be discussed.

There are three basic elements of virtual memory. Virtual address space, the main memory, and the auxiliary memory (Carr, 1984).

The address space is the set space, which is reserved for the use of memory swap. It creates a distinct location for data to be held. This process tremendously speeds up the retrieval time of virtual data. To the task, the address space is simply a continuous array of words or bytes, each with unique virtual address. To the virtual memory hardware, the address space is partitioned equal size frame or blocks (Carr, 1984). The auxiliary memory has a much longer access time but a lot less expensive than main memory. Conceptually, if one thinks of the program in the secondary memory as the original copy and the pieces brought in to the main memory as copies, rather the other way around (Tanenbaum, 1976).

Keeping the original program updated must not be over looked. When a program needs a page out of virtual memory, the operating system must find where that page is located. This task is simplified by partitioning the drive to allow virtual memory to be stored in a specific part of the hard drive.

The pages that are placed in frames inside virtual memory are known as the task.s resident set. The resident set pages and the frames in which they reside is identified by a data structure called the Page table (Carr, 1984). The User can remove one or many pages out of the resident set. As a program executes, it accesses sequence virtual addresses in the address space. If the page containing a virtual address is resident in main memory, the information at the virtual address is properly obtained or updated.

When a program references a page and then cannot find it in main memory, this creates what is called a .Page Fault.. After a page fault has occurred it is required for the operating system to read in the required pages from the secondary, or virtual memory. Then the operating system must enter its new physical memory location in the page table, and then repeat the instructions (developer, web). If the operating system cannot complete the task the system will crash. This is also known as .Fatal Page Fault..

Address mapping device is the essential hardware element in virtual memory. It is located between the processor and the main memory. This device uses as page table to translate each virtual address in the address space, to a real address in main memory (Carr, 1984). The address-mapping device would interrupt execution if the page were missing. It would then signal the operating system for a page fault. Two activity indicators exist in the main memory frame, a use-bit, and a dirty-bit. The bits are in a separate memory and have one pair for each main memory frame. The used-bit indicates that the page is active and, most likely, should not be replaced. The dirty-bit indicates that the page must be moved to an auxiliary memory before it can be replaced (Carr, 1984). When a missing page is referenced the operating system creates the page table for each task. The system, then, selects a main memory frame to hold the missing page and performs an I/O operation to move the page from its slot to the frame. The frames that have not been allocated can be chosen. All the frames are physically identical except for its real address. This brings us to the page replacement algorithm. There are many page replacement algorithm policies, such as; LFU (Least Frequently Used), LRU (Least Recently Used), and Random. The most effective page replacement algorithm policy is the LRU. It will replace the memory slot being used the least amount of time.

A problem that can be created by virtual memory is when the overall supply of memory is running low then swapping may occur. If the memory