ns2 server

Hescominsoon

Hescominsoon

CGA\TOJ Hosting Manager
with my wife's extended unemployment i'm now effectively working three jobs..running my business..working at a tire shop..and a landscaping company. is the server still being used? How is it performing..etc etc.
 
Uh yeah, we play on thursday nights. Sorry for forgetting to reply.

We'll probably kick the slots down from 14 to 12 again. 14 doesn't really work well.
 
glad to hear it. I am working on a couple of upgrades for the host that will greatly improve the working of the ns2 virtual machine. first is maxing out the hose ram(that won't help the ns2 box but wil help me)..then i'm going to get a hardware raid card so i can convert to vmware. What vmware will allow me to do is pool all of the ghz of the host cpu(about 8.8 ghz counting vmware overhead) and then i can setup a burst pool for things like ns2 to use. I can reflectively give ns2 4-6 ghz of bursting power so additional slots will be usable(providing overall load is low and it usually is)..:) that should mean 14 slots won't be a problem. it's all about getting about $400 together..:)
 
I think upstream bandwidth, rather than processing power, is the main issue. When we have 6-8 players, the servers runs just fine. It holds solid at 10 player but when 11 or more players are on, we see noticeably higher latency and rubber banding.

I'm not saying the processing power upgrades won't help, but all evidence suggests bandwidth is what's keeping us from moving beyond 12 slots of reliable performance.
 
bandwidth is fine...the server has never even gotten to 50% of my outbound bandwidth..ns2 is very cpu bound..they want 4ghz i ahve right now 2.4ghz i can dedicate to ns2 due to the limitations on how hyper-v allocates cpu. the vmware changeover will allow me to throw up to 5 ghz at ns2 if needed..:)
 
here's the bandwidth stats for the interface the ns2 server resides on. I have 10 megabits up..not up to..a hard 10 up:

[TABLE="width: 100%"]
[TR]
[TD="align: left"]Interface Usage Sprint Femto (eth2) (inbound)[/TD]
[TD="align: right"]928.34 kbps[/TD]
[TD="align: right"]31.00 bps[/TD]
[TD="align: right"]4.17 kbps[/TD]
[TD="align: right"]59.00 bps[/TD]
[/TR]
[TR]
[TD="align: left"]
Interface Usage Sprint Femto (eth2) (outbound)[/TD]
[TD="align: right"]18.56 Mbps[/TD]
[TD="align: right"]0.00 bps[/TD]
[TD="align: right"]16.44 kbps[/TD]
[TD="align: right"]8.00 bps

[/TD]
[/TR]
[/TABLE]
for my firewall on this particular interface outbound is to the served network and inbound is from the served network so effectively the terms are relaly revresed..it's strange but that is how they see it. so the maximum of data this interface saw(and this includes the ns2 night) was 930 Kbps..i have 10mbps..:)

if you'd like i can post the full weekly and monthly chart..:)
 
Huh. I had just assumed that the higher latency after more than 10 people join was related to bandwidth, but it seems I was incorrect.

Sounds like we'll be able to bump the server to 16 slots without issue once you can allocate more processing power to the server, which will be awesome! :)
 
yeah i wish hyper-v would just pool the mhz like vmware does..but ms thinks it knows better..and they don't..:)
 
You must log in or register to reply here.
Back
Top