Back when we were figuring out where to go after Eve, @joshuafcole and I built some systems for various companies to see if there was a good fit. We never showed them publicly, but the one we did for @stripe Sigma was really cool:
https://www.youtube.com/watch?v=tSk3_ujnu14 …
cc: @patrickc
-
Prikaži ovu nit
-
Thinking through how Sigma's infrastructure must work, we guessed that it had to take all the normalized information about your transactions and denormalize it into SQL rows that can get dumped into some big analytic database. From their site, that process took 48 hours.
1 reply 0 proslijeđenih tweetova 2 korisnika označavaju da im se sviđaPrikaži ovu nit -
Using
@rustlang we built a version of that system that could run in real time while maintaining high transaction throughput. The video is the visualization of the throughput of that system and a little SQL runner to show it hitting Redshift.1 proslijeđeni tweet 6 korisnika označava da im se sviđaPrikaži ovu nit -
We did some back of the napkin math for what lyft's transaction rate would be, which is one of Stripe's biggest public customers. Our estimate was that it's in the 10-20 txn/s range, but might burst to 10x that at times.
1 reply 0 proslijeđenih tweetova 2 korisnika označavaju da im se sviđaPrikaži ovu nit -
We also looked into Visa's global transaction rate, which was around 1200-1600 txn/s. The demo you're seeing in the video is us handling 10x Visa's rate on a laptop. On a mid-tier AWS instance, we got up to 10x the transaction volume of the world :)
0 proslijeđenih tweetova 15 korisnika označava da im se sviđaPrikaži ovu nit
The visualization was my favorite part, it was based on liquid motion timers, which I really loved as a kid.pic.twitter.com/qC1dQD7FgO
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.