Here's How I Scaled A Cryptocurrency Exchange's Trading Engine to 1 Million TPS

Written by siddharth | Published 2020/07/31
Tech Story Tags: cryptocurrency | golang | concurrency | scaling | crypto-exchange | cryptoexchange | cryptocurrency-exchange | crypto-exchange-architecture | web-monetization

TLDR The matching engine takes in multiple orders at once, maintains them, does the computations and pushes out trades. The server crashed consuming 100% CPU capacity, messaging queues started getting stuck. A single failure or any wrong calculation could put the exchange at risk, mostly at the financial end breaking the user's trust. Go-routines have minimal context overhead. They do context switching only in well-defined situations, go-routine becomes co-operative, very likely to become very likely.via the TL;DR App

no story

Written by siddharth | I imagine, design and build interesting ideas.
Published by HackerNoon on 2020/07/31