Round robin explained as a work-splitting trick you can use in real life
This article was originally written in July 2010 on my old blog, mixture-art.net, and has been moved here. The writing style and information are kept as they were at the time.
Isn’t that a cool-sounding word, “round robin”? Hmm, sounds like Robin Mask’s brother or something.
Well, round robin is actually a term from communications (IT). Say “communications” and it sounds intimidating, but to put it simply:
A round robin is a method of dividing up work. (Starting to feel like something you could use in real life?)
Let’s start with a bit of background. In the world of communications, lots of servers and even more clients are constantly exchanging all kinds of information moment by moment, and that’s what makes everything run.
For example:
A certain client: “I want to do a Google search” A certain nearby server: “Ah, in that case you can connect to the hogehoge server. I’ll let them know.” Server hogehoge: “Ah, got it, I’ll connect to Google… ah, the results are in. Forwarding now.” A certain nearby server: “Got the results back from hogehoge, sending them your way.” A certain client: “Received, thank you! ♪”
All of this happens at insane speed. Just thinking about it is exhausting.
The example above had only one client (meaning “customer”), but in reality it’s nothing like that. With tons and tons of clients saying this and that, do this, do that, what happened to the thing I asked for earlier! — the server is working frantically every day just to process all that information.
So, right before it bursts, what do you do?
What would a human do? Divide up the work, right?
So, in human society, say there’s “a document that has to be made by tomorrow.”
A certain customer (client): “Could you get that explanatory material ready by tomorrow?” A certain guy (server): “Yes, got it. (Ugh… easy to say, but there’s a lot of it…)” A certain coworker (?): “Oh, that looks rough. Want a hand?” A certain guy: “Thanks, you’re a lifesaver~” A certain coworker: “Then let’s call in my junior too and split it three ways. Hold on, let me call him…” A certain junior (?): “Hello?” A certain coworker: “Hey, some work came up, want to help out?” A certain junior: “Huh? Seriously? I’ve got a date with my girlfriend today, so today’s rough…” A certain coworker: “C’mon, don’t say that, we’re supposed to help each other out when things get tough.” A certain junior: “No, seriously it’s rough, and what makes it worse is she gets really scary when she’s mad, seriously.” A certain coworker: “I don’t care about that, just do what you can, what time’s the date?” A certain junior: “Seriously, just today, please cut me some slack~” A certain coworker: “You know…” A certain guy: “Ah… it looks like a hassle, so… it’s fine if you can’t help…” A certain coworker: “That’s not the point, it’s this guy’s whole attitude toward work that—”
Yikes, this doesn’t look like it’s going anywhere. Would’ve been better to just do it alone? If they’d secured the junior’s schedule ahead of time, or decided in advance that whatever work comes to any of the three gets split three ways — if they’d set a rule for how to divide things up beforehand, none of this would’ve happened.
One such rule is the round robin.
Regardless of who’s free or whose work is faster, the work is simply divided equally, and in preparation for jobs that keep piling up, there’s also a rule like “work on this task for X minutes, and if it’s not done, set it aside for now.” Rather than sitting around debating which has priority, just get moving. And so the whole thing doesn’t stall because of one job that’s dragging on, if it’s not done within a set time, it just gets pushed back for now.
In the sense of mechanically streamlining things without overthinking, it’s a pretty well-suited way of dividing work. Not a bad rule for automation.
In the networking world there’s “DNS round robin,” which makes it look like there’s only one server when there are actually several, and there’s also “weighted round robin,” which adjusts how much work each of several virtual servers gets when their processing speeds are already known to differ.
“Priority of work” and “optimal division of labor” are really difficult things, aren’t they? In times like that, maybe it’s fine to just go with a round robin.
Originally published in Japanese at https://clazytech.com/2010/07/1971/. Translated with LLM assistance and reviewed before publication.