MongoDB documents take up a specified amount of space on a disk. If we perform an update that increases the size of a document, this may end up being moved out of sequence to the end of the storage block, creating a hole in storage, resulting in increased execution times for this update, and possibly missing it from running queries. The compact operation will defragment space and result in less space being used.

We can update a document by adding an extra 10 bytes, showing how it will be moved to the end of the storage block, and creating an empty space in the physical storage:

compact can also take a paddingFactor argument as follows:

> db.runCommand ( { compact: '<collection>', paddingFactor: 2.0 } )

paddingFactor is the preallocated space in each document that ranges from 1.0 (that is, no padding, which is the default value) to 4.0 for calculating the padding of 300 bytes for each 100 bytes of document space that is needed when we initially insert it.

Adding padding can help alleviate the problem of updates moving documents around, at the expense of more disk space being needed for each document when created. By padding each document, we are allocating more space for it that will prevent it from being moved to the end of the storage block if our updated document can still fit in the preallocated storage space.