Computers of today multitask where you can run more than one program simultaneously. All computers need system memory to operate and require a method of management to insure the proper handling of the data stored within it. Memory management involves relocation, protection, sharing, logical organization, and physical organization of the data. Without proper memory management page faults would climb out of control and data could become corrupted or lost.
As a process waits for its turn with the CPU the operating system may move the process out of the main system memory into the page file and back again into main memory. When it is moved around it may not be in the same memory address as it was before it was relocated. This is where memory management comes into play as it helps the operating system translates the memory references found in the processes code to the physical address of the program in the main system memory. Processes also need the protection from other processes for interfering with its operation. In this case process cannot access or reference memory addresses without permission. This is handled by the computers processor as the operate system cannot predict every memory reference that all running programs will make (Stallings, 2009).
Sharing of system memory is another technique used by memory management. If you have many processes accessing the same program, allowing all processes to access the same memory address instead of requiring each process to have its own memory space will free up system memory. The part where memory management comes into play here is giving each process a turn with the memory address location (Stallings, 2009).

Finally we have the origination of memory whether it is Logical or Physical organization. Logical origination deals with the linear origination of main system memory in a computer. This linear origination does not correspond with the way programs are written and is the reason memory management is need. Physical organization deals with the fact that computer memory is organized in to two different levels, main memory and secondary memory. Main memory cannot provide permanent storage but is fast, where secondary memory provides permanent storage, is large in size, but slow. The transfer between main and secondary memory requires memory management to insure that running process are utilizing the faster main memory while insure all data is preserved on secondary memory (Stallings, 2009).
Replacement algorithms are methods to transfer data between virtual memory and main system memory. There are different types of algorithms such as first in first out (FIFO) and last recently used (LRU) that attempt to decrease the number of page faults. A page fault is when the computer needs to transfer a page from main memory to virtual memory in order to make room for the needed page that is currently located in virtual memory (Vt.edu, 2012). When observing the java applet displaying the 5 different algorithms used to predict what pages to move in order to decrease the total number of page faults, they all tend to be close in percentage rate of around 50%, See figure 1.1 below. This shows that the unpredictability of the processes running on a computer make it difficult to predict what the next process will require from memory (Vt.edu, 2012). Figure 1.1

Virtual memory provides the operating system with more memory for all running processes. With the size of programs increasing the amount of main system memory has also increases, but still can hold all the needed data for all running processes. When a process is running it will access the needed data from the main system memory as it is much quicker the secondary memory that resides on the computer’s hard drive. The operating system will chose different pages of system memory to store in a page file also known as virtual memory thus freeing up the needed, faster system memory for the current running process.
Another function of virtual memory happens to deal with computer errors and memory dumps. Since system memory does not store any data when the system is powered off, the computer will dump the system memory into virtual memory if an error occurs that will require the computer to shut down. This insures that data is not lost in the case of an error. References
Stallings, W. (2009). Operating Systems: Internals and Design Principles. Sixth Edition.
Vt.edu. (2012). Virtual Memory. Retrieved from http://courses.cs.vt.edu/~csonline/OS/Lessons/VirtualMemory/index.html