Hacker News new | ask | show | jobs
by cd34 5155 days ago
Companies that speak about certainty and their server being the fastest, need to take the time to make sure it is.

http://simonhf.wordpress.com/2012/04/26/g-wan-versus-sxe-hel... is a benchmark showing sxe, using gwan's benchmark code, as outperforming gwan 1.8x.

Years ago, there were two very fast servers - Zeus and Tux. Tux was a kernel mode accelerator that avoided context switches and Zeus was the de-facto standard of 'fastest userspace webserver'.

Chromium X-15 came along but skipped a few compatibility features, and was labelled 'faster than tux, but in userspace'. Tornado did something similar, very narrow purpose, but, as they started to bolt on all of the pesky RFC stuff, became a top-midline application server.

While GWAN is certainly fast, but not fastest at running its own benchmark, their boasts will ultimately affect public opinion. Language is a barrier - perhaps some of their boastful attitude is merely rough translation.

GWAN's use case basically works around almost every webserver's dream - more cores = odd setups to take advantage of those cores through cpu/irq affinity, etc. GWAN handles that out of the box which is a definite advantage.

If you needed an app to do some calculations and hand back results with the least hardware possible, GWAN would be a top contender. Hardware is still fairly inexpensive that it would take a rather large company that would be able to take true advantage of the cost savings of reducing their hardware outlay based on GWAN's scaling.