Having come from a time where loading a small JPEG required a progress bar, it seems like one beefy server machine ought to be able to handle thousands of concurrent requests per second. This seems like it should fit most companies' needs until a hundred million in revenue or so:
1000 req/sec * 6 hours of daytime in one market * 365 days = 1.3 billion, so $0.10 revenue/req would be $130 million. Seems like even node.js could do 1000 req/sec unless the database bottleneck is large, let alone something like Go.
Of course, I might be wrong about this, these are just unverified estimates.
But, another reason is that I, personally, don't want to manage computers, I want to do interesting things with code. As soon as you get a second computer, infrastructure starts becoming non-trivial. I am not interested in infrastructure, so I'd like to be sure I've maxed out one computer first.
Also, my set of problems that I'm interested in does not really require distributed systems.
Comments
Having come from a time where loading a small JPEG required a progress bar, it seems like one beefy server machine ought to be able to handle thousands of concurrent requests per second. This seems like it should fit most companies' needs until a hundred million in revenue or so:
1000 req/sec * 6 hours of daytime in one market * 365 days = 1.3 billion, so $0.10 revenue/req would be $130 million. Seems like even node.js could do 1000 req/sec unless the database bottleneck is large, let alone something like Go.
Of course, I might be wrong about this, these are just unverified estimates.
But, another reason is that I, personally, don't want to manage computers, I want to do interesting things with code. As soon as you get a second computer, infrastructure starts becoming non-trivial. I am not interested in infrastructure, so I'd like to be sure I've maxed out one computer first.
Also, my set of problems that I'm interested in does not really require distributed systems.