Networking FreeBSD x Linux

I'm not impressed.

For starters he's using FreeBSD STABLE which is a developer snapshot that has plenty of debugging code in it, so obviously it'll be slower. He doesn't bother to provide any arguments as to why he did that so I can only assume that he assumes that STABLE means what it says, which it doesn't.

I also couldn't help but grin when he started talking about using his own coded tool to measure some performance, where it immediately becomes clear that he had to adjust the code to run on FreeBSD (see the Github page), definitely no bias involved there for sure :D

All in all I can't take this very seriously.
 
srlemke
It's a brain dead article by the clueless.
Source: former employee of mine who worked at Netflix and is now at Google. Networking is something I wish I could spend time on cause I love fundamentals. I can't so I hire or talk to experts in the field, as high up as I can go. I don't get a chance to talk to my friend much anymore but, coincidentally, we crossed paths online when that article came out and that is what he said to me.
 
In general: any new user posting here with a post that is essentially "read this interesting/fun/controversial link" is just clickbait.
I have the hope that new users quickly learn that posting relevant and interesting content here is much better.
 
For starters he's using FreeBSD STABLE
The article mentions "FreeBSD 11.2-RELEASE" so he probably updated it in the meantime. The differences in performance are quite substantial. I would like to see better arguments than "it's a brain dead article by the clueless" to counter the tests done by him. I do not believe the difference can be that big but where's the flaw in his tests and setup then? Can someone from the BSD team perform a better comparison?
 
I would like to see better arguments than "it's a brain dead article by the clueless" to counter the tests done by him. I do not believe the difference can be that big but where's the flaw in his tests and setup then?
First of all you're responding to a thread which is almost half a year old, kinda late to the party.

And second I already mentioned this in my original post: the guy did his measurements using a tool he wrote himself instead of using commonly known (and trusted) performance measurement tools. If that doesn't ring any bells then I don't know what will :rolleyes:
 
He also used commonly known tools: "As a first test, iperf3 and netperf were used."

Sometimes you just have to write your own tools, take, for example, the authors of iperf and netperf.
 
Back
Top