Collection of useful, relevant or just fun places on the Internets for 18th March 2012 and a bit commentary about what I’ve found interesting about them:
Performance Calendar » Introducing mod_spdy, a SPDY module for the Apache HTTP server – Not many people are using Apache these days because it’s too slow, but still this is a good start on getting SPDY adoption.
mod_spdy is still in early beta, and is not yet recommended for deployment in production environments. If you’d like to test out mod_spdy and help us to make it better, please consult our Getting Started guide. We hope to make it production-ready sometime in early 2012. Stay tuned by subscribing to our discussion forum.
draft-mbelshe-httpbis-spdy-00 – SPDY Protocol – Draft IETF standard for comment and review.
Jumbo Frames on vSphere 5 « Long White Virtual Clouds – Another field test with Jumbo frames points out that a 10% improvement is well worth doing.
A 10% performance degradation might not sound like much, but when you’re talking about a 10Gb/s network that’s like losing the performance of an entire 1Gb/s link. When you use multiple links it quickly adds up to be a substantial loss of performance. The benefit of Jumbo Frames is only going to grow with the new 40G and 100G Ethernet standards. Let’s just hope that the OS IP stacks are improved enough to cope with the new standards when they start to become mainstream.
Understanding EZVPN Authentication – Astorino Networks – Great article from Joe on EZVPN authentication.
draft-ietf-tcpm-initcwnd-00 – Increasing TCP – The Google sponsored standard to increase the TCP initial window to 10 improve performance over the Internet.
This document proposes an optional standard to allow TCP’s initial
window to start at 10 segments or roughly 15KB, updating RFC 3390. It discusses the motivation, the advantages and disadvantages of the higher initial window, and includes test results from several large scale experiments showing improved latency across
the board for a variety of BW, RTT, and BDP classes.
Why upgrading your Linux Kernel will make your customers much happier – This article talks about tuning the Linux TCP slow start alogorithm to imrpove performance. I would normally use an F5 load balancer with TCP One Connect which is much more intelligent than this – but hey, if your only tool is a hammer, every problem looks like a nail. Nice summary though.
Translated into figures, the Linux box with the IW of 10 is 21 percent faster when you look at total time. If you discount the connection round trip it is about 25 percent faster. All of this without a single change to the application.