About a week ago an interesting question was posed to the members of our Enterprise Systems Group on LinkeIn and I want to share that question along with two very interesting responses in this Blog.
Here’s the question that was posed by Tonya Trotter: “What’s the best way to explain a mainframe to children?”
There have been a number of terrific responses but I thought this one provided by Monte Bauman was especially creative and interesting.
“To use an analogy kids can "get". I explain that mainframes are like cities, for example New York City. And other computers are like towns (like Mayberry).
In the cities and towns are factories, shops, and homes, where work is done (analogous to CPUs); there are also parking lots, loading docks, storerooms, closets, and shelves where stuff is stored either before or after work is done on the stuff (analogous to data and memory and cache levels); and cities and towns have warehouses and storage units where stuff is stored for long terms (analogous to disk/tape); and the cities and towns are connected by highways, roads, airports, and waterways (analogous to networks).
In a "mainframe" city there are numerous large factories with large warehouses and large parking lots and lots of storerooms and big closets and wide shelves where stuff is moved and stuff is made (analogous to batch workloads); and there are numerous shops with large lots of big shelves where stuff goes to be sold (analogous to transaction processing); and there are many, many homes with lots of closet space where stuff is used (analogous to online processing). A "mainframe" city is interconnected with numerous wide streets that go from factory to factory and store to store with but few stop lights that are well-timed and dynamically controlled to optimize rapid movement through the city (analogous to the MCM).
In a "mainframe" city, there may well be a downtown and a number of suburbs (or boroughs) that compose the city, where each section of the city may take on a specialization (analogous to LPARs ... dev test prod). The movement of "stuff" from factory to factory and shop to shop is tightly controlled in each suburb to optimize the movement of "stuff" and when conflicts occur "priorities" are enforced to make sure the most important "stuff" is done first (analogous to WLM).
When conflicts occur between suburbs in a "mainframe" city the issues are rapidly resolved according to shared policies and priorities (PR/SM and IRLM).
The entire "mainframe" city and all factories and shops and homes, downtown and suburbs, are monitored with fine-grained granularity and administered from a single building run by the mayor (z/OS) and his helpers, the business manager (WLM admin), the police chief (RACF admin), the warehouse manager (storage admin), and the highway supervisor (network admin).
A town? (aka a non-mainframe server). Much the same can be said for a town, except that a town likely has just a single factory and a couple of stores and some homes and a single ramp off the interstate. To get a lot of "stuff" done in towns requires a lot of towns and a lot of dependance on the highways and a lot of cooperation between the mayors of those towns. And we know how well government officials like to cooperate (they all have their own priorities now don't they). The Police force in a town is not nearly as sophisticated as a city’s and thus security exposures exist in towns that do not exist in cities (think CSI vs. Andy in Mayberry). When a lot of "stuff" needs done all at once in a town there are but few resources available to bring to bear, even if a governance mechanism existed to request such a reorganization of resources, which there isn't.
So where do you want your "stuff"? Where do you want to get your "stuff" done?”
A somewhat simpler explanation was just provided by Dave Sienkiewicz whan he posted the following: "Kids, it's like having two strong oxen doing the work of a hundred hamsters".
Both explanations work fine for me.