First, I understand that the schematic was of a universal design for maximum flexibility. We used to call those type of schematics "Microcomputers" as opposed to "Microcontrollers"; the later being a cleverly minimized I/O design.

It doesn't represents a typical, good 8051 design. I shudder to think that anyone would base a new product design based upon that example without without thinking. Note that the microprocessor in such a Microcomputer design could just as well have been a Z-80, 8080, or whatever you choose to use. By the time you add that much external circuitry the costs all balance out wherein the only deciding factor in microprocessor selection is programmer's preference.

The power of the 8051 is its ablility to fill the niche between the insect-world PIC processors and a generic microcomputer design. The Zilog Z-8 can do this as can Motorola's insane cornecopia of psychodelic microcontrollers; the 8051 does it best.
It gives you processing power, fantasic interrupt handling in a package that supports minimal external requirements.

I typical 8051 design I do uses no external memory of any kind and certainly no external 8255. I'd be ashamed to stick an 8255 in one of my designs as it suggests to me that the design wasn't minimized properly to reduce cost and complexity. I think 8255 are only used by college instrutors to help them standardized their microprocessor lectures.

Here's my design progression heuristic:
=======================================
(1) PIC - Used for minimal I/O, minimal processing.
(2) Philips 8051 - Minimal I/O, good interrupt and processing.
(3) Anything Else - Here it doesn't matter because the cost of all the chips for external suppot tend to even out no matter who's micro you use. Rather like the Central Limit Theory in statistics, with all the external parts and their random count&costs, it all balances out along a normal distribution.

-Jay C. Box