Code and servers pitfalls to avoid when launching social application with many interactions?

Viewed 71

Assuming you want to launch a social app (which mean many interactions) with the ambition of acquiring several thousand users and for those who have already done so, what are the pitfalls that you know and that you would absolutely avoid, in term of code and servers architectures for sure ?

I have that feeling that you can easily feel alone when you trying to answer this kind of question which is clearly out of the scope of all that SaaS or landing pages that maybe (and I insist on this word) don't have this scaling problem. Or maybe that there is just no real pitfall and that the best approach is 'problem' --> 'solution' when those problems come up.

I don't think it is an opinion-based question because I/O intensive database, queue systems, server calculation, etc have clearly some technical consideration in that kind of configuration.

And to give you some example of problems that I think large scale social app can encounter, there is Facebook engineers with their early latency problem or Twitter engineers with their Bieber problem.

I was able to avoid the first pitfall that Netflix couldn't avoid, which is not using Cloud and trying to build their own servers infrastructure at this scale.

1 Answers

The CAP theorem is vital for all large distributed systems. It is explained here.

Related