I hereby confirm Fresh 1.4's perf-improvements are legit!
Fresh 1.1
❯ wrk -t12 -c400 -d10s http://www.russbrooks.com
Running 10s test @ http://www.russbrooks.com
12 threads and 400 connections
Requests/sec: 121.80 👈
Fresh 1.4.2
❯ wrk -t8 -c500 -d10s http://www.russbrooks.com
Running 10s test @ http://www.russbrooks.com
8 threads and 500 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 27.93ms 38.66ms 514.91ms 93.96%
Req/Sec 1.34k 0.98k 3.60k 65.40%
105725 requests in 10.07s, 19.96MB read
Socket errors: connect 253, read 0, write 0, timeout 0
Requests/sec: 10495.36 👈 🚀 🚀 🚀 🚀
That jives with the numbers here, where Fresh placed second-fastest framework.
Lighthouse Score
Equally incredible!
Hosted on a free account at Deno Deploy. Imagine what the numbers would be on Enterprise-class infra. This means we finally have Fire-and-Forget websites, because a single Node is able to handle a 10X more traffic than your Ruby/Node/Python ones. No more scrambling to scale every 8 weeks, after your team launches 1.0, having to answer pages at 3 AM because your Scripting-Language just collapsed at 500 R/s, then again at 1,000 2 mo later, then 1,500 2 mos later, then 2,000, then 2,500, etc.
Our Rust boxes never trigger PagerDuty.
I'm really pleased with the whole Deno experience, from toolset, to perf, to pkg mgmnt, to sensible defaults, to the TypeScript Std Lib that's bundled, to the Edge-First global deployment you get free.
Top comments (0)