Hacker News new | ask | show | jobs
by CoolGuySteve 1724 days ago
"Sorry. Unless you’ve got a time machine, that content is unavailable."

Too bad, it would be nice to see someone go through and document how Twitch works. I've never worked at "web scale" so I'd probably learn a lot.

4 comments

> I've never worked at "web scale" so I'd probably learn a lot.

As someone who has worked at both large and small companies, you'd probably be disappointed.

It's likely lots of bubble gum and chicken wire. I'm sure in the video ingest and transcode side of things there are some really interesting bits though. When you're owned by Amazon you don't need to optimize too much to achieve web scale... just leverage AWS services. It's not like you're going to get a bill.
> When you're owned by Amazon you don't need to optimize too much to achieve web scale... just leverage AWS services. It's not like you're going to get a bill.

Oh you're be surprised. Divisions get billed constantly for the AWS resources they consume, and this bill gets taken out of their annual budget. From what I hear, this is a common practice in most large organizations.

Also, the AWS services you can access from within Amazon are almost identical to the AWS services you can access as an external customer. It's equally easy/hard for a random company to achieve web scale, compared to Twitch.

A lot of it is probably hacked together -- like, embarrassingly hacked together lol
This is true about almost any company. Closed source generally means you can have lower standards.
You’re being downvoted for being overly negative, but the ops code is of (literally) shockingly poor quality.

This leak has made me understand clearly that code quality is not what makes a product great.

I guess that’s something.

The jenkinsfiles are mostly nice and clean though. I’ve definitely seen worse of those.

Oops, didn't mean to be too too negative. I say embarrassing in the sense of, I've definitely shoved out awful code because something needed to get out(tm). And with large companies, deadlines that cause that situation are inevitable.

But I also say it like that because, well, I've seen code that causes (objectively easy-to-fix) crashes but still ships because of one reason or another: laziness, politics, inexperience. It's a part of software engineering I'm still trying to accept.

Yep, there are lots of small services that don't seem production ready in the source code. Though admittedly we don't know which of those are deprecated.
Well, you know what they say, "Self help is the best help."
I hear Netflix has a good tech blog ;)