Computers have been so fast for a good + 15yrs. If you don't use bloatware and pile on layers and layers of crap and with good system engineering it's amazing the amount of work you can compute for so little cost.
> I would also like to meet this guy, buy him a beer/coffee and thank him for the value he has created:).
I had the chance to met him a few years back during a conference in France[1]. He is very friendly indeed.
1: https://mixitconf.org/en/2015/thibault-duplessis-lichess-org...
There's a replay of this talk (from 2015) here: https://www.infoq.com/fr/presentations/lichessorg-open-sourc...
Reminds me memories :)
I had the opposite reaction.
$1000/mo on a few frontend servers for only 20K concurrent users seems excessive.
I've seen simple $40/mo droplets handle 50,000 concurrent websocket connections with minimal latency without breaking a sweat.
$2000/mo on databases is also nuts, unless you're storing hundreds of terabytes at high redundancy levels.
> $2000/mo on databases is also nuts, unless you're storing hundreds of terabytes at high redundancy levels.
What? 300TB at 0,20e /GB which would actually be quite cheap for fast storage at high redundancy would cost 60 000e / month.
1 billion chess games per year, often with >20k concurrent users and and some very heavy-duty engine analysis occurring for some proportion of these games and they only spends $56k.
I have seen startups with only hundreds of customers spend 10x this on AWS by copying the fancy tools used by tech giants.
It is also interesting how little the creator pays himself ($56k) and how much he spends on data protection ($40k), taxes ($52k) and various administration tasks. I wonder if his time is as heavily skewed towards administration as his expenditure.
I would also like to meet this guy, buy him a beer/coffee and thank him for the value he has created:).