Impossible Engineering Problems Often Aren't | Hacker News Reader

Impossible Engineering Problems Often Aren't

blog.scalyr.com

53 pointsspiffytech11 comments

Be VERY careful with this sort of advice: "Infrequent user actions can lead to complex server actions."

Malicious users use that knowledge to craft disproportionally powerful DOS attacks. Your site is as vulnerable as the most CPU-expensive feature.

At the very least, rate limit those features.

I guess I read "infrequent" differently: anything a user can do will inherently be frequent given enough scale. On the other hand, that "export everything ever to CSV" button in the admin panel? That can take a few CPU seconds. It won't matter, because only five people even have the ability to hit that API endpoint, let alone the inclination.

If you enjoy reading articles about the rediscovery of indexing large amounts of read-only data, I'd highly recommend reading this book which is a treasure trove about this kind of work:

http://www.amazon.com/Managing-Gigabytes-Compressing-Multime...

I was sort of disappointed that this wasn't about actually impossible engineering problems [1].

[1] https://en.wikipedia.org/wiki/Atlantropa

(Author here)

I can see that the title comes off as too grandiose out of context. Sorry about that; I was looking at it from a different angle when I wrote the article. The idea is that in normal work, you often run into challenges that seem infeasible. This should be a trigger to look for a way to reframe the problem, rather than giving up.

I wasn't trying to bust your balls, really, just poking a little fun at the hyperbole. I enjoyed the article for what it was.

Yeah, building a space elevator, or a practical SSTO space shuttle plane. Handling server monitoring hardly seems worthy of such a title. http://en.wikipedia.org/wiki/Space_elevator http://en.wikipedia.org/wiki/Single-stage-to-orbit

Yeah I thought we're gonna talk about things like space elevators, or other megaengineering projets.

http://en.wikipedia.org/wiki/Megascale_engineering

"Reframing the problem" is often easier in the virtual world. When someone does come along with a good reframing in the physical world, the results can sometimes be very dramatic though.

Sadly, I'm slowly getting used to such disappointment. There's this other thread that's #1 now on the front page, with a title: "Terraform". Guess what's it about. Tip: not planetary engineering.

You're free to write the post you wish to see.

Reply on news.ycombinator.com