Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Does it scale? Who cares (2011)

jacquesmattheij.com

Post reply on HN
Print view View original

171–180 of 285 posts

Re: Does it scale? Who cares (2011)

#171

Post by frik »

narrator wrote: post #87
technosamay20 wrote: post #15

Couldn't agree with this article more. I built the biggest social network to come out of India from 2006-2009. It was like Twitter but over text messaging. At it's peak it had 50M+ users and sent 1B+ text messages in a day. When I started, the app was on a single machine. I didn't know a lot about databases and scaling. Didn't even know what database indexes are and what are their benefits. Just built the basic produ…

>I know of absolutely no service which failed because it couldn't scale. These days there would be hardly any way to fail to scale given a large budget. The technology to do so is all so good. However, back in the day it wasn't always so. I'd say Friendster was an example of a company that failed due to scaling issues.

Some things simply don't scale. You can't do some things in real-time. You have to do it in batch-jobs or avoid them. If you know about the O-notation, you will probably know certain algos don't scale. Friendster insisted to calculate the friends-of-friends-of-friends relationship in real-time and was not able to scale it. MySpace and Facebook did the same calc in the early days too, but both scrapped the feature as they were not able to scale it. Other examples are big data (e.g. log analysis), if you choose the wrong software stack (database) or wrong indexes you won't be able to scale it (especially if it's real time. You will bleed money in no time for cloud servers with TBs of RAM, when you could do it with a couple of servers with a normal sharded SQL database and one hand tuned index.

Re: Does it scale? Who cares (2011)

#172

Post by nyir »

Dylan16807 wrote: post #130
tbrownaw wrote: post #18

Earlier quoted context omitted.

Microsoft Access. Works fine for something one team uses, works... not fine for something the entire corporation uses.

Does Access actually work well for 20 users?

Anecdotally it doesn't: had to use a time tracking program that used an Access database on a shared network folder and would frequently freeze up or crash due to locking issues (with around that many users). I imagine it could be done better though.

Re: Does it scale? Who cares (2011)

#173

Post by hopfog »

weitzj wrote: post #136
hopfog wrote: post #122

I'm in the unfortunate position where this question actually matters from day 1. I learnt the hard way a few days ago when I hit the bottleneck (about 50-100 concurrent users) and I'm not sure how to proceed. It's a multiplayer drawing site built with Node.js/socket.io. I'm already on the biggest Heroku dyno my budget can allow and it's too big of a task to rewrite the back end to support load balancing (and I wouldn…

Maybe you could rely on Firebase for the real-time db stuff and do the drawing there.

I actually used Firebase for the first version but ran into a lot of different issues, mostly due to the lack of control. That's why I migrated to my own Node server where I could have my own back end logic to control things.

This was many years ago though so maybe I should have a look at it again.

Re: Does it scale? Who cares (2011)

#174

Post by hopfog »

Everhusk wrote: post #126
hopfog wrote: post #122

I'm in the unfortunate position where this question actually matters from day 1. I learnt the hard way a few days ago when I hit the bottleneck (about 50-100 concurrent users) and I'm not sure how to proceed. It's a multiplayer drawing site built with Node.js/socket.io. I'm already on the biggest Heroku dyno my budget can allow and it's too big of a task to rewrite the back end to support load balancing (and I wouldn…

Hey, you might want to check out https://realm.io/ , the demo is exactly what you said. Good luck!

Looks cool but seems like it's aimed at mobile and not websites? Also, how do they differ from Firebase?

My previous experience with Firebase makes me hesitant to use a BaaS. It works great for simple stuff but as soon as you want more control it becomes a mess compared to just having your own server. It might just be me who haven't wrapped my head around the mindset correctly though.

Re: Does it scale? Who cares (2011)

#175

Post by hopfog »

stevewilhelm wrote: post #140
hopfog wrote: post #122

I'm in the unfortunate position where this question actually matters from day 1. I learnt the hard way a few days ago when I hit the bottleneck (about 50-100 concurrent users) and I'm not sure how to proceed. It's a multiplayer drawing site built with Node.js/socket.io. I'm already on the biggest Heroku dyno my budget can allow and it's too big of a task to rewrite the back end to support load balancing (and I wouldn…

I think what the author of the original article would suggest that you need to convert some of the concurrent users to paying customers and use the money to purchase additional Heroku dynos.

I don't think that's possible for me. I have some future monetization plans but that requires features I don't have yet. At the current state I have a hard time seeing anyone would pay for it. Also, due to how the back end is structured it's not possible to scale horisontally at the moment so all I could do with money is beefing up the current dyno, not adding more.

Re: Does it scale? Who cares (2011)

#176

Post by hopfog »

true_religion wrote: post #135
hopfog wrote: post #122

I'm in the unfortunate position where this question actually matters from day 1. I learnt the hard way a few days ago when I hit the bottleneck (about 50-100 concurrent users) and I'm not sure how to proceed. It's a multiplayer drawing site built with Node.js/socket.io. I'm already on the biggest Heroku dyno my budget can allow and it's too big of a task to rewrite the back end to support load balancing (and I wouldn…

Heroku is fairly expensive. Could you not port this to a more dedicated server? Yes, it's likely complicates deployment but for a few hours of work in initial deployment, you can save hundreds of hours rewriting your codebase.

I'm thinking about it but I'm no server admin so it feels a bit scary. I could probably do it quite easily but it would be a security nightmare.

Re: Does it scale? Who cares (2011)

#177

Post by hopfog »

yxhuvud wrote: post #129
hopfog wrote: post #122

I'm in the unfortunate position where this question actually matters from day 1. I learnt the hard way a few days ago when I hit the bottleneck (about 50-100 concurrent users) and I'm not sure how to proceed. It's a multiplayer drawing site built with Node.js/socket.io. I'm already on the biggest Heroku dyno my budget can allow and it's too big of a task to rewrite the back end to support load balancing (and I wouldn…

What is it your processor spends its time on? Perhaps those parts could be rethought and redesigned? Perhaps heroku is the wrong solution for those tasks even if it is good for the rest of the app. Would it be possible to find something cheaper for just those parts?

To be honest I haven't really benchmarked it. Maybe that's something I should invest time in to see if there's some obvious bottleneck I can fix. Thanks for the suggestion.

Re: Does it scale? Who cares (2011)

#178

Post by hopfog »

spacetexas wrote: post #146
hopfog wrote: post #122

I'm in the unfortunate position where this question actually matters from day 1. I learnt the hard way a few days ago when I hit the bottleneck (about 50-100 concurrent users) and I'm not sure how to proceed. It's a multiplayer drawing site built with Node.js/socket.io. I'm already on the biggest Heroku dyno my budget can allow and it's too big of a task to rewrite the back end to support load balancing (and I wouldn…

Socket.io can't handle many connections you're right. I would use (universal) Web Sockets (This is a good abstraction library, very easy to implement). https://github.com/primus/primus Maybe put it on a Digitalocean Server and just reference your Dyno's IP address if possible.

Thanks! Looks good so that's definitely something I should look more into.

Re: Does it scale? Who cares (2011)

#179

Post by baby »

technosamay20 wrote: post #15

Couldn't agree with this article more. I built the biggest social network to come out of India from 2006-2009. It was like Twitter but over text messaging. At it's peak it had 50M+ users and sent 1B+ text messages in a day. When I started, the app was on a single machine. I didn't know a lot about databases and scaling. Didn't even know what database indexes are and what are their benefits. Just built the basic produ…

How much money did Twitter waste in its migration to Java? I think it's an important factor to consider.

Re: Does it scale? Who cares (2011)

#180

Post by hasenj »

qume wrote: post #154
technosamay20 wrote: post #15

Couldn't agree with this article more. I built the biggest social network to come out of India from 2006-2009. It was like Twitter but over text messaging. At it's peak it had 50M+ users and sent 1B+ text messages in a day. When I started, the app was on a single machine. I didn't know a lot about databases and scaling. Didn't even know what database indexes are and what are their benefits. Just built the basic produ…

I built a service (~10mm users at peak) which was designed from the ground up to scale, around 2002. When the numbers did grow we just sat back and watched it pretty much. Even given this, I completely agree with you. That's why I now develop in ruby. I'll take developer productivity over performance any day. It's the old saying - 'nice problem to have'.

> That's why I now develop in ruby. I'll take developer productivity over performance any day.

In my experience, dynamically typed languages don't do well for developer productivity.

Post reply on HN