![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Geeky Stuff
Aug. 6th, 2002 05:55 pmOnce upon a time, there was a Danish mathematician named Agner K. Erlang. He worked in the telephone field in Europe, in the 1900s and 1910s. Erlang solved one of the thorniest mathematical telephone problems of the time: how many phone lines should be available at the phone company. Anyone, at any given time, should be able to pick up the phone and get a dialtone. However, the phone company doesn't want to buy the equipment to provide dialtone to every subscriber simultaneously when, most of the time, the phone is sitting there unused. They want to buy just enough equipment to guarantee that there are enough dialtones for whoever might happen to pick up the phone. It's a probability thing. And, since people don't call 24 hours a day, Erlang's math solved the problem of how many dialtones were probably necessary during the phone company's "busiest hour". This "busy hour" traffic unit is now called, of course, the "Erlang," and Erlangs are very common calculations in my industry.
I hate Erlang.
In my narrow, niche part of the telephone industry, I commonly have to figure out how many phone lines my clients must buy. They may want one of those phone trees ("Press 1 for sales, 2 for support, 3 for any human voice for godsake...") and they want to make sure they have enough phone lines for anyone who may call in at any given time. Erlang calculations are always used, and they're always wrong. A business, for example, usually has two slightly smaller busy hours instead of one very busy hour, one starting at 10:45 or so, and the next starting at 1:30. Before and after lunch. Erlang is too high, and the client has bought too many lines. A client once sold pre-paid phone cards to Mexican bodegas; calls from the US to Mexico were scattered and random Monday through Saturday, then peaked unbelievably and unnaturally on Sundays from noon to three. Erlang drastically underestimated the desire for Mexican card users to call their mothers after church.
I'm gonna go entrepreneur again. I'm writing my business plan, something I dislike only slightly less than Erlang calculations. One of the things I have to work out, for myself this time, is how many phone lines I need to buy for my service. *sigh*
I hate Erlang.
In my narrow, niche part of the telephone industry, I commonly have to figure out how many phone lines my clients must buy. They may want one of those phone trees ("Press 1 for sales, 2 for support, 3 for any human voice for godsake...") and they want to make sure they have enough phone lines for anyone who may call in at any given time. Erlang calculations are always used, and they're always wrong. A business, for example, usually has two slightly smaller busy hours instead of one very busy hour, one starting at 10:45 or so, and the next starting at 1:30. Before and after lunch. Erlang is too high, and the client has bought too many lines. A client once sold pre-paid phone cards to Mexican bodegas; calls from the US to Mexico were scattered and random Monday through Saturday, then peaked unbelievably and unnaturally on Sundays from noon to three. Erlang drastically underestimated the desire for Mexican card users to call their mothers after church.
I'm gonna go entrepreneur again. I'm writing my business plan, something I dislike only slightly less than Erlang calculations. One of the things I have to work out, for myself this time, is how many phone lines I need to buy for my service. *sigh*
