Partitioning

Fixed Partitions

One of the simplest allocation methods divides memory into fixed-size partitions, each holding one process. When a partition is free, a process no larger than the partition can be selected from the input queue. After termination, another process can use that partition.

Fixed partitions are simple and have low overhead, but their fixed sizes and count prevent programs larger than the largest partition from running and limit the number of active processes. Internal fragmentation reduces memory utilization.

Regardless of its size, a process occupies a whole partition. Unused space inside that partition cannot be used by another process: this is internal fragmentation.

Dynamic Partitions

When a new process needs memory, find a sufficiently large block and allocate the required space if one exists.

Three common approaches to dynamic storage allocation are:

1. First-fit: start at the beginning of the list and allocate the first sufficiently large block.
2. Best-fit: search the entire list and allocate the smallest sufficient block. This minimizes the immediate remainder, not necessarily long-term external fragmentation.
3. Worst-fit: search the entire list and allocate the largest block, leaving the largest immediate remainder.

Dynamic partitioning allocates suitably sized regions and eliminates internal fragmentation in an ideal exact-allocation model, but creates external fragmentation. Loading and removing processes divides free memory into small pieces. External fragmentation occurs when total free space can satisfy a request but no sufficiently large contiguous region exists.

Compaction consolidates separate free partitions into a larger one. Moving relocatable processes to one end of memory brings free regions together at the other end.