What is composition? I know a poorly constructed photograph when I see it; I know a poorly constructed sentence, too. (That one, for instance.) But what is composition?
I once mentioned to a potentially great Flickr lass that her photos needed a keener eye to their composition. No picture frame or plant sticking out of her head, for instance. It's a common problem with snapshot-style photographs; the concentration is on the subject, and no one actually sees that sign post, or the stalk of that "plant that adds a bit of drama" as it sprouts from the unfortunate noggin.
Composition can encompass things like angles that point in disparate directions. Sometimes those disparate angles can help us understand the picture; they help us distinguish what is important, and what is not. Sort of like a movie that has a subplot; the diversions help us understand the whole thing. (I'll assume the movie is well-written.) More often, disparate angles are exactly that - disparate. They fight each other.
In hospital, I was invited to join in a small art class. For that class we had to draw a circle (plates were provided), and then draw and paint our image inside that circle. It suited me, perfectly. I was going to draw the view from my bed - but every horizontal line would have a different vanishing point. Those points I figured out mathematically. (It took a few moments; I didn't have a calculator. No, I'm not gifted with numbers, but I do "see" spatial relationships.) Each point varied by "e", the natural logarithm. I never did finish that picture. I was called to answer questions by the local Gestapo representative Social Worker. (Hat tip to Renée for the linguistic technique! :-) ) I was well ahead of her. More when I feel like it.
I tried to finish that picture, a few years later. I still have it, in my mind. It will never be finished; or if it is, I'm back there, or in somewhere similar. (Never happen. Never.)
When I look at a network diagram, I can immediately pick up if it's any good. I know a good network when I see it. I don't need negative space to help me pick out the salient points. I look at a good network as a piece of art - if it's good, I want to hang its diagram on my wall. If it's bad, I want to ask "Why?"
Once upon a time I had to figure out a network. The designer was a highly paid, quite influential, consultant. He decreed that all network resources should be equally spaced from the every resource. ie, Every server had to be an equal number of hops from every desktop.
How communist. How stupid.
His dream machine didn't take into account the speed of light, nor the speed of a signal through a fiber-optic cable. My refutation did; it was simple: I merely stated that his network would fail because the odd number of "master" nodes had a resonant frequency that equalled a "stalled network". That is, a network that would oscillate with itself trying to figure out who was the master.
Look up the FDDI specification and see if you can figure out, without mathematics, the problem. A clue: it's at 7 nodes. In fact, it turns up at all prime numbered nodes, above 7. (Prove it? Who am I? If you want mathematical proof, try and interest Zoe Brain; she's far cleverer than I!) Doubt it? It's fairly simple: the time it takes a system to figure out who's the master, the next round of figuring out who's the master has begun. All you need to know is the average speed of light through a fiber, and a guess at how quickly the router can answer.
I still cringe at the proof.
Oh, the network well enough. It even stayed up. But what I came to call the "harmonics" - the secondary signs - were out of whack. And than an apprentice electrician cut the cable. He was supposed to. No one but one or two people knew what they cable contained: the primary and backup networks. Neat, yeah?
(I later came across the same oblivion when someone mirrored a 2GB disk: he dissected it in half, and made one half mirror the other half. Oh yeah, that's a mirrored disk.)
I remember when the trading network went down. I had questioned the wisdom of putting " "those" servers in a different building to the trading floors they served. But it all server the desire to minimize the "hop" count. That buzzword was all the rage among the clueless.
Where does art come into this? You know, I've spent a hell a long time trying to figure that out. I know that when I was given the task of designing an investment bank's international network, the first place I went was MOMA. If memory serves, it often doesn't, I even took a day off. I studied paintings, sculptures how light changed what we see. It was quite a wonderful time.
I designed a network that was supposed to last 18 months. It was still in place 5 years later. It was well overdue for replacement. It was a poor successor who decreed that it should last longer than it was designed to. (Opinionated? Me? Funny you should ask... :-) ) Oh, I'm sure if you ask certain individuals my design was destined for the scrap heap before I put pen to paper. It was a pen, too. Visio didn't exist. Beside which, I used Visio to formalize my pen and paper musings.
We were sat on a red eye from California; I think from LA, but it could have been SFO. At the time, I think I took the flight to the Golden State more often than I took the cross-town bus. The chap next to me was a movie producer; who, I have no idea. I didn't ask his name. He spent a the first few hours of the flight figuring out his organization for a movie he was working on. He had some fancy Windows program to do it with. I had another international network - 15 locations, 9 countries - to design, based on tests I'd (designed and) conducted, and meetings held. I had a sophisticated, for its time, laptop. I also had a tablet of paper and a decent pen. (I'd bought it from a very nice chap in the airport; for some reason, I recall the purchase, but not the airport.) LA? Nooo... I don't think it was LA. Orange County? Perhaps. It's not important. Where was I? Oh yes. He used an expensive program to design an org chart. I used a pen and paper to design an international network.
I still think about that moment.
My early career was spent actually handling wires. And the voltages they can carry. I have perfect color vision, so I was given the task of learning how to join subtly colored wires together. Funny story: I knew a chap who was color blind; he figured it all out from where the wire was in its cable. Then they changed the color scheme, and how the wires were twisted together in the cable. Apparently it took "them" a long time to figure out his vision problems. (Is it, in contemporary vernacular and standards, to re-iterate this story? Is it "ablist" in some dismal, mediocre way to remember that chap with affection, and still laugh at his plight? If it is, sod ya. Go buy a sense of humor, because it's the only way you'll get one.)
Ah fuck it. I'm going to go all post-modern, and not actually define a conclusion. I haven't finished designing networks. I still know what a good network design is. I feel it's not the crap so many produce.
Want to know what a good network is? Go to a good museum with lots of modern art. Figure out what the artist is asking. And then go figure out your network. If the questions don't align with the answer - go do some more work.
Good network design does not rely on protocols. That's like a poet relying device, a painter relying a specific light or a fool relying on the right answer to magically appear. It never happens, and when it does - you know it. Network design is one of the new crafts.
Carolyn Ann
