New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
As in deciding which poison will act fastest and be the less painful?
Do we know that the Russians or Chinese spy or poison site? And by "know" I don't mean wild media or politicians allegations but proof.
We seem to have proof for the Chinese government filtering and limiting their surfers and we seem to have proof that the Russians blackhole or otherwise block (access to) sites they deem dangerous or criminal. But so do many if not most governments.
I know of only 2 governments who have been proven to spy and hack - and neither speaks chinese or russian; both speak english.
Btw, my point is not pro or anti this or that government. My point is more how almost obscenely things are bent. If a "good" government really does something bad , no matter how evil, it's somehow OK (e.g. "necessary"), but if a "bad" government is alleged of doing something bad, just alleged, the media shout it out day and night (and blow it out of proportion).
Other than this sentence, I think you kind of went back to whatever your point was, but this still confuses me. The poison site is usually at the place of injection, FWIW.
Seems like they left you out of the loop. China 2025 domination is an open secret. As well, 40 year plans have leaked over the last few years talking about taking over as #1 world power. They are executing a long term strategy and don't change their top guys like US does every 8 years. You can easily see their business practices and patent abuses to destroy western leaders.
Seriously, when you downplay it that much, you sound like you're intentionally trying to mislead.
If nginx goes away, sure there will be enough forks.
FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK FUCK!
One decent web server and they FUCKED it up.
Fucked into damnation.
The end!
@deank Knew it all along.
Well then, big fat
sausagesapache2 for all!https://www.nginx.com/blog/nginx-joins-f5/
"I’m incredibly excited that today we announced NGINX has signed a definitive agreement to be acquired by F5."My ass.
You are incredibly excited because you got $670m which was a pocket change to the chinese!
Here is the incredibly excited ass:
https://www.nginx.com/people/gus-robertson/
Quite similar.
$670m for an event loop :facepalm:
plus a logo
The best laid plans of mice and men often go awry.
Edit1: typo
Edit2: I am sure US is begging China to stop the trade war. Also, I probably have way more PRC pals as data points.
Unfortunately not true for speed at least, both caddy and h2o (my own custom binaries and official) are slower in terms of throughput and latency than nginx (at least my centmin mod built nginx) at high levels of concurrency 1,000-10,000 users. But better in some ways yes, h2o has proper HTTP/2 request prioritisation implementation of which nginx is not 100% on out of the box - though you can tune nginx to perform better for initial page and start render times. You can test each web server's implementation of HTTP/2 prioritisation by reading https://github.com/pmeenan/http2priorities/tree/master/stand-alone. Edit: with that said I can see why some folks might notice h2o loading pages faster over HTTP/2 HTTPS due to out of the box generic nginx configs not having a good HTTP/2 request prioritisation implementation compared to h2o.
Last I tested also both caddy and h2o performance noticeably drops as you add more http headers to the response compared to nginx. So when you test compare nginx and h2o/caddy remember to try to use the same amount of http response headers. Though it's been alot longer since I tested h2o.
FYI, I build my own h2o and caddy binaries for testing integration with Centmin Mod and those h2o/caddy binaries are already faster than regular versions i.e. my own caddy binaries are 20-40% faster than official caddy ones but still slower than my centmin mod built nginx binaries (centmin mod nginx is 2-3x times faster in throughput and 2-5x times faster in latency response times than custom caddy built binaries for HTTP/2 HTTPS TLS 1.3 load tests). Benchmarks at https://community.centminmod.com/threads/caddy-0-11-5-tls-1-3-http-2-https-benchmarks-part-1.16779/. Granted my custom built binary for caddy 0.11.5 with go 1.12 is first caddy build that has properly handled high user concurrency HTTP/2 HTTPS load tests unlike prior versions.
@eva2000
Well, it depends on how far one wants to go with that. What about,for example, fcgi-speed?
It may be good or bad but I approach those things from 3 points, (a) security, (b) as a developer, and (c) as an admin.
For (a) there isn't much to say. Nothing (beyond a couple of lines) in C or C++ - or on Go, for that mattert - can offer a solid level of security. It's just technically not possible. But one can at least try hard, which is why I highly value e.g. the OpenBSD people.But there are some (still very young) languages like e.g. Zig that might allow us to write a secure server that's also performant and, that's important, that are relatively easy to use and that will actually be used and not only by stubborn security freaks (like Ada).
For (b) I see a very major and decisive point that biases me strongly in favour of H2O. Let me explain: H2O already has got the difficult parts right. Performance will merely be a question of optimization; sometimes simple, e.g. by replacing a slower hash with a faster one and sometimes more tricky, but still just optimization.
nginx, on the other hand, has already been heavily optimized. There's not that much breathing room left. Plus, more importantly, nginx still has to get the hard part (protocol with all its corners and traps) right. Based on what I see there I predict that H2O will overtake nginx in speed.
For (c) the points go to nginx at least for the foreseeable future. It takes years to build lots of communities, hundreds of howtos, guides, tutorials, etc.
And frankly, I'm not sure that we should care a lot about speed (in that region). Both will have at least very good speed anyway and I don't see a lot of sense in hunting the top speed between two candidates that are in the leading group anyway.
My take is that people like myself who are mainly interested in a http/2 front end/proxy will tend towards H2O while thousands and thousands of happy nginx users will stay with nginx (or its coming major fork). And btw, if there is a contender for ngnix then it's caddy from what I see and not H2O which doesn't have the hype behind it that Go/caddy have.
Indeed this is a big factor in determining what is the right server for each individual user .. but also for upstream users too i.e. more likely web hosts/tech support/system management companies/folks will have a better understanding of ins and outs security/troubleshooting wise of apache and nginx, then they would of h2o and caddy.
It is interesting indeed, it's why I follow and test different web servers with the intention of integrating it into my Centmin Mod stack installer. Right now it's nginx, but eventually will support apache 2.4, openlitespeed, litespeed enterprise and either or both h2o/caddy but performance for me a big deciding factor as to what will be integrated over time.
still alive ... amazing? how have you been? i hope all is good from your side. welcome back.
The creator of nginx was born in Kazakhstan, Almaty. So maybe he is a Kazakh spy...
Also, creators of Yandex (2 jews) studied there.
I attended the same school where he studied
"Russian", "Kazakhs", "Jews"? Huh???
I don't care at all. My only bias in that regard is that Russians (and probably Kazakhs) usually have good school/university education.
My experience working with Russians (and presumably Kazakhs) was always excellent. They had a very good education and were disciplined thinkers (and usually very friendly persons). Besides that purely professional perspective I do not care at all about "Russian", "Chinese", "Kazakh", "Jew", "buddhist", etc. etc. The creator of H2O is a Japanese afaik; I don't care.
I prefer my webservers pillow-laced by grandma. In grandma I trust.
@ehab: All well, thank you very much! Hope the same for you!
jews
@eol
THAT is hilarious! Too funny.... Thank you, I needed a good laugh.
...this subject took a different track....
@Amitz Welcome hope you are doing well.
This is a lovely bunch of coconuts. Ahh well all good things come to end I suppose.
https://www.lowendtalk.com/discussion/comment/2952488#Comment_2952488
We miss the green 🍄 hope he may come back
Yeah.
@eva2000 have you tested tengine? What's your opinion?
Meh.
You need to compile from source http://tengine.taobao.org
I know.
My point is that it's not in the official repositories.
Which sucks from free software point of view, but more importantly, it sucks from practical point of view - these deployed servers need to be maintained. Having something standardized makes deployment, maintenance and scalability easy.
Plus, i believe in leaving systems usable and easy manageable for other admins too.
Hacking and screwing around available code is always fun, but compiling and packaging is a job for software developers.