Five Koans of Software Architecture

https://bellmar.medium.com/five-koans-of-software-architecture-f9f7305598c2

by mbellotti • 4 years ago

25 14 4 years ago

bellmar.medium.com

tgv2• 4 years ago

are you going to be on call for this

When does a call at 3am involve an architectural decision or change? It should never come that far.

teddyh• 4 years ago

It’s a variant of the old “Would you bet your career on this?”:

https://dilbert.com/strip/1992-09-06

jboy55• 4 years ago

I've been the victim of this, a producer consumer system with a single index file using some variant of a heap. For each incoming request the producer; locked the file, read the heap from the file, added the value to the heap, then wrote it back and unlocked it. For each consumer thread, the thread did the reverse. It was setup with threads created by incoming requests, and a threadpool of 128 threads for the consumer.

This worked seamlessly until one day, at 3am PST, a flood of requests came in and the index file grew to ~4mb, and everything ground to a halt.

The legend of the architecture was a senior engineer did this on a weekend, then a month later left the company.

yokaze• 4 years ago

I think, you are misreading it.

It isn't about making an architectural design decision / change at 3am, but someone has to live with the consequences of such a design decision and fix something in some way or another at 3am in the morning.

I.e. you could go with a clustered db setup which is supposedly HA, or a classical single instance DB...

The point here is, whoever has to get up at 3am to fix a potential catastrophic failure in the setup should have a say in it.

tgv2• 4 years ago

If the architect is also the coder/support, then it's a moot point. But if there is a separate architect, or a pack of them, there's no point in waking them up at 3am. Any architectural change isn't going to be deployed within the hour.

Personally, I think the designers/architects should work on the code too. You have to dogfood your own design, so to speak.

waynesonfire• 4 years ago

I agree, if you design it right, which isn't easy.

BurritoAlPastor• 4 years ago

When did “koan” become a synonym for “aphorism”?

bugbuddy• 4 years ago

When everyone started believing in themselves as part-time Zen masters after watching a TED talk video on YouTube.

tombh• 4 years ago

That's the real koan.

xtiansimon• 4 years ago

I agree. Koan is supposed to be strange mind worm that contradicts your normal experience, sound of one hand clapping sort of thing.

Come to think of it, the AI peeps might be able to make a few. They stay stuff like, does a submarine swim? Pretty close.

Kalanos• 4 years ago

graph dbs are just easily traversable joins of all many-to-many tables. although they have way less tooling in most ecosystems to support them (e.g. REST API was tough to do), they are extremely powerful [realizing i should write a post]. so, i would only use the entities i was modeling were natural graphs: chemistry, supply chain, network.

neomodel is a fantastic python ORM for a neo4j: https://neomodel.readthedocs.io/en/latest/

AndyPatterson• 4 years ago

Build it the stupid way first is short sighted imo but this might just be an exaggerated way of saying premature optimisation is wasteful.

Just putting it out there, but I think Knuth’s aphorism is being used and abused to the point that some folk believe system qualities are second to functional requirements when it’s not the case.

bkuehl• 4 years ago

I'm probably a bit late on this, but none of these points touched on a software engineer building a complex architecture for no reason. I have seen this so many, many times. Yes, they can handle point #1. They take all the tickets associated with it and solve them quickly because they built it and know all the connections. The implementation makes no attempt (or claim) at scalability so that's a moot point. For point #3, interestingly enough these are never about optimization of anything, but about abstraction over abstraction over abstraction. As for the graph DBs, those aren't allowed anyway by the company they work for (thank goodness) or they would find some way to incorporate them.

Yes, I'm being a bit glib, but over-engineering does not always mean there's some good intentions in there. It can just be really bad.

snapdaddy• 4 years ago

People in the comments are so quick to be snarky, but this is all really good advice.

Five Koans of Software Architecture

Related Stories