Hacker News Reader

Comment by focom | Hacker News Reader

What would be the equivalent in software engineering? Loosing your domain name?

When using AWS for anything, pray for the health of us-east-01. When that region has problems the whole internet quakes with fear. You never know what's going to break.

I'm shocked at how true this is. A huge amount of software is only redundant by being hosted in mutiple parts of aws us east.

That's the funny part. It is spread around 'US East'. If anything catastrophic (or mundane) knocks out power/internet in that region it won't matter how many backups you have. It will bring down large parts of the internet.

That seems to indicate a problem with the decision making behind putting important things in only one region.

I'm happy if it goes down , nice excuse to not get any work done that day

I hear "bus factor" bandied about.

https://en.m.wikipedia.org/wiki/Bus_factor

Prod database on a single server with a single drive

...with no backups.

Without reliable restores.

FIFY

What part of "with no backups" didn't you understand?

:-)

Perhaps the part where "having backups which are unreliable/unrestorable" is _more_ dangerous than "being known to have no backups", because the former provides a false sense of security?

"Good companies test their backups. Great companies test their restores."

Learned that lesson the hard way a few decades ago.

And stored under the developer's desk.

with the power button lined up with where their shoe rests.

A CEO once got up on stage to announce an important new phone to the world's assembled press not knowing that the picture on the screen of the phone was coming from a PC in a bedroom closet and the developers were sweating like dogs as they watched the broadcast on the BBC. I shall say no more.

Who needs backup when you just rsync every 10 minutes?

Rsync?

MTBF on a WD Green is 1M hours, we’ll replace it at 900K and be golden…

An interesting question, what in software causes catastrophic issues when it's broke, but also designed as it should, like the Jesus nut is for a helicopter?

Obviously many systems have a single point of failure, and can result in lost productivity, but is that designed as it should, or cause catastrophic results like a helicopter crash?

Load balancer? Cluster coordinator? STONITH device

BGP

Jesus Undersea Cable?

OT: that's how i recognise French people anywhere ;) only one o for "losing"

Huh? That's an idiomatic error typical of native speakers. People who learn English as a second language tend to do much better on the "loose-lose" and "there-their-they're" kinds of errors

For what it's worth, in my experience the GP's comment rings true. I don't see many native English speakers making that mistake, and I do see lots of native French speakers making it.

> People who learn English as a second language tend to do much better on the "loose-lose" and "there-their-they're" kinds of errors

This sounds logical, but I am not sure about it. I certainly make much much more errors based on similar sounding words in English than on my native language. (I believe I make a TON of such errors in English, and almost none in my native language, but hard to tell.)

I can confirm that as a non native English speaker I don't make this kind of mistake often. I think spelling is easier for me because I first learned how the word is spelled and only later how to actually pronounce it? And i separate homonyms like they're/their in my head because they're very different words in my language.

Idiomatic error is "would of"

haha good catch

DNS

Encrypting your data at rest and losing the key.

Reply on news.ycombinator.com