How HAProxy forwards 2 Million Requests Per Second? - The Backend Engineering Show

The Backend Engineering Show with Hussein Nasser

English - May 10, 2021 19:27 - 47 minutes - 27.3 MB - ★★★★★ - 5 ratings
Technology Homepage Download Apple Podcasts Google Podcasts Overcast Castro Pocket Casts RSS feed

Previous Episode: The Tale of OLTP, OLAP, and HTAP in Data Warehousing - The Backend Engineering Show with Hussein Nasser

Next Episode: This is why Salesforce services went down on May 11 2021

In this show, I go into detail on how HAProxy achieved 2 million HTTP requests per second. This is a very well-written article that discusses how the HAProxy team benchmarked the product on a 64 core ARM machine leading to over 2 million requests per second. There are many components and low-level points that I try to elaborate on, timestamps below.

0:00 Intro

2:40 Summary of the Article

11:55 Latency and Throughput in HAProxy 2.3 vs 2.4

21:00 How TCP Connections Affects Performance

28:00 Maximum Packets we can get in 100Gbps Network?

35:00 How 64 Cores are divided between workloads

40:00 Tail latencies HAProxy 2.3 vs 2.4

42:50 How TLS Affects Performance?

HAProxy Blog https://www.haproxy.com/blog/haproxy-forwards-over-2-million-http-requests-per-second-on-a-single-aws-arm-instance/

Support my work on PayPal

https://bit.ly/33ENps4

Become a Member on YouTube

https://www.youtube.com/channel/UC_ML5xP23TOWKUcc-oAE_Eg/join

🧑‍🏫 Courses I Teach

https://husseinnasser.com/courses