Live data from Hacker News

Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

infoq.com

41–50 of 114 posts

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#41
post #36

Earlier quoted context omitted.

I laughed. Deft trolling. Bravo.

Why? Looks like a legitimate question to me.

Perhaps so, but then it would fit into the category of 'unaware of being a parody of itself' statements/questions that are also (unintentionally) humorous such as the classic "I can tell that site was built in rails from the design".

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#43
post #16

Earlier quoted context omitted.

http://bugmenot.com/view/infoq.com Bugmenot is great for sites like this.

http://10minutemail.com/ is what i use, much simpler.

I wouldn't call that simpler. You have to register using a fake email, possibly actually go check the inbox and click a link, and possibly be required to log in again now that the account is registered. With bugmenot, all you do is copy and paste a valid username and password.

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#44

Earlier quoted context omitted.

Last time they shared, they were spending ridiculous amounts of money on Amazon ec2 a month. Like ridiculous amounts. Someday, the money will run out, and they'll have to try and turn a profit. Their business model is fundamentally bad.

If those outdated figures of ~$50k/month are true, I wonder if they could move to their own infrastructure or dedicated and turn some of those savings into getting from red to black.

Yes.

We operate a dozen of our own colos, with a virtual colo on AWS for insta-scalable multi-region redundancy, and an Amazon "colo" costs the same as about eight of our own when spun up and serving at least a gigabit of traffic.

However, the difference is less if you're going from zero sys admins to 24/7 says admins. I'd SWAG the crossover is once your AWS budget exceeds 4 full time sys admins willing to do shift work.

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#45

Thrilled to finally see this on InfoQ -- already an underappreciated resource for technical talks. What is it about video + slides that appeals so little to people?

What is it about video + slides that appeals so little to people?

A full transcript with interleaved slide images takes a few minutes to read at most, and lets you control the pace of information absorption. A video with slides, especially when you cannot 2x the talking speed, is a painfully slow data transfer method.

Video + Slides = analog modem

Transcript w/ embedded slides = Google fiber :-)

Of course, if you're into audio books instead of reading, maybe you consider that a feature.

// I push video for a living. It's great for visual explanation like DIY instruction (e.g. woodworking, swapping RAM on a Mac Mini), emotional content, personal story telling, etc. Systems architecture is generally not in one of those categories.

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#46
post #4

What a horrible website. Made an account to get the mp3. Just to discover that the mp3 download link doesnt work even then. Thank god I used a throwaway email to sign up. Otherwise I would be even more pissed.

http://bugmenot.com/view/infoq.com Bugmenot is great for sites like this.

i keep forgetting about bugmenot

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#47
post #14

Earlier quoted context omitted.

If there is one lesson that HN needs to learn, it's that profitable is not the same thing as important.

So they are important, but need to live off ramen every day (figuratively). What good does that do to them? I don't think grandparent should be downvoted, he raises a good point. Tons and tons of people use Reddit, but Reddit has a hard time making a living.

> So they are important, but need to live off ramen every day (figuratively). What good does that do to them?

Maybe they enjoy it? Maybe it makes them happy?

Not everything is about money.

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#48
post #7

I am curious how much a Go lang rewrite would make a difference in scaling up.

It depends how much time is spent doing computation and how much time is spent doing lookups from disk. If the latter is clearly dominating, then a switch to Go will not help much.

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#49
post #16

Earlier quoted context omitted.

http://bugmenot.com/view/infoq.com Bugmenot is great for sites like this.

http://10minutemail.com/ is what i use, much simpler.

For Firefox users, made even smoother with: https://addons.mozilla.org/en-us/firefox/addon/bloody-viking...

Re: Scaling Reddit from 1 Million to 1 Billion – Pitfalls and Lessons [video]

#50
post #36

Earlier quoted context omitted.

I laughed. Deft trolling. Bravo.

Why? Looks like a legitimate question to me.

Some time ago the HN hivemind went through a period of blind Node.js love. Now it's going through blind Go love.

Regardless of what you think of either language, "rewrite in X" is not a magic incantation that will spontaneously solve all your architectural issues. Designing a good architecture involves balancing many components, of which your primary implementation language is an important, but not exclusive, element. There are also the organisational issues -- hiring, spending time not adding new features, etc.

Post reply on HN