As I recall the concept of an address space arrived with the advent of virtual memory. At least that's when I first became aware of the term.
Since the internal "main" memory required by a pgm no longer corresponded to its size the term "address space" was coined to identify the space required bt the progam.
Although the "main" memory required by a pgm is now a fraction of its load module size it still is assigned a region that reflects the overall size of the pgm.
This region now exists on DASD (virtual storage) rather than real (main) memory and the segments of the pgm are paged in to real memory. The number of pages present in real memory at any given moment is called the working set.
Well, that's what I recall. There may be (most likely) some minor misstatements and omissions, but it should give you a ppretty good overview of the concept.
An address space is the amount of virtual storage seen by a job
MB in MVS/SP and older systems
GB in MVS/XA and MVS/ESA
Each job executes in an address space
Online users each have a Time Sharing User (TSU) address space
Batch jobs execute in an Initiator address space
An address space is organized as a set of:
Segments (1MB each)
Pages (256 pages per segment, each page is 4K)
An address space is either:
Swapped out (all segments are on auxiliary storage)
Swapped in (at least one segment is in real storage)