Expected Server configuration for 100 users

[Scott Larson]

     I can second the fact FreeBSD + nginx is a rocking combo. We've been
running that for years under ever increasing traffic and it only requires a
few basic adjustments to the OS, even fewer in 10 since a lot of system
defaults were cranked up for modern times. Our current hardware handling
nginx related duties are sporting Intel L5520 procs and have been in
service since 2009, only now in the process of being replaced due to aging.
If systems had a guaranteed infinite life these boxes would still be good
for quite some time. If you go the Linux path it's probably a similar
situation, nginx is likely not going to be the part of your stack which
cracks first.




*__________________Scott LarsonSystems AdministratorWiredrive/LA310 823
8238 ext. 1106310 943 2078 faxwww.wiredrive.com
<http://www.wiredrive.com/>www.twitter.com/wiredrive
<http://www.twitter.com/wiredrive>www.facebook.com/wiredrive
<http://www.wiredrive.com/facebook>*

On Wed, Feb 18, 2015 at 11:01 AM, Rainer Duffner <rainer at ultra-secure.de>
wrote:

>
> > Am 18.02.2015 um 16:56 schrieb ragavd <nginx-forum at nginx.us>:
> >
> > Hi,
> > We are configuring the NGINX as a reverse proxy. We are expecting some
> 100
> > concurrent users or connections/sessions to be active at any given
> moment of
> > time. Right now the server is acting as a reverse proxy for only one
> > application. These concurrent users will connect predominantly between
> 6:00
> > AM to 7:00 PM. Based on this what should be the RAM and CPU configuration
> > for the NGINX server?
> >
>
>
> What’s your hardware?
>
> It all depends on a couple of configuration-parameters.
> http://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_buffers
>
> If I’m correct, the default would be
> 8*4k=32k per connection, which would result in a memory usage of 3.2MB
> with 100 concurrent connections.
> Of course, nginx itself would also need some memory.
> But nginx is quite thrifty IME.
>
> > Also is there a guideline or a blog entry which we can use to approximate
> > the system requirements of NGINX servers based on the concurrent user
> load?
>
>
> I use the above formula as a guideline.
>
> But I haven’t really had a situation where I was ever coming close to
> hitting a limit on the hardware or our (very modest) worker_connections
> default.
> If we get DDoSed, it’s usually so much crap-traffic that we have to route
> the affected network through a mitigation-service.
>
>
> > And also what will be preferred OS for NGINX server?
>
>
> I think the Tier 1 platforms are:
>  - FreeBSD 9+10 AMD64
>  - Cent OS 6+7 AMD64
>  - Ubuntu 12+14 AMD64
>
> NGINX Plus supports a couple of additional platforms, but I would assume
> the majority of the installation-base is on any of these three (and then
> some Debian installs).
>
> http://nginx.com/products/technical-specs/
>
> For your use-case, it doesn’t really matter what OS you run, as long as
> it’s one of the above (or you know it really well).
>
> I think FreeBSD9+10 have the lowest hardware requirements, even without
> special tuning.
>
>
>
> _______________________________________________
> nginx mailing list
> nginx at nginx.org
> http://mailman.nginx.org/mailman/listinfo/nginx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nginx.org/pipermail/nginx/attachments/20150218/1bef5829/attachment.html>