From: cciestudy (cciestudy@mid-world.net)
Date: Tue Jan 20 2009 - 13:13:18 ARST
Each device is tied to a specific telnet port and you can only telnet to
each device one at a time. It is a matter of how many devices you can run
and what the performance is like.
In my experience with dynamips these are the factors that effect performance
and the number of devices you can run:
1) server CPU. More the better.
2) Somewhat server memory, less these days. Less when using the same IOS
version.
3) number of dynamips instances. I usually use 3-4 devices per dynamips
process. If you put all devices in one dynamips process it will not work
well.
4) dynamips IOS idlepc value. This is IOS version specific and needs to be
"figured out". If this is calculated wrong performance will be terrible.
5) 64bit linux with 64bit dynamips. In my testing linux blows away windows
for dynamips performance.
As an example, I have setup a training server for my current customer. It
is a 4 way dual core server. 8 3ghz processors total, 4gb RAM, 64bit fedora.
I can run 33 2725's, with BGP, ATM, frame and 1 VMWARE instance and have it
usable for training. During peak convergence I get 80-90% cpu utilization
on all i8 processors.
Hope this helps.
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Supreme Nkosi
Sent: Tuesday, January 20, 2009 4:37 AM
To: ccielab@groupstudy.com
Subject: Multiple Dynamips Sessions on a shared server
Please advise.
Just wanted to find out if anybody has ever set up dynamips on a
linux/windows
server and had two or more people log on simultaneously using any remote
client
(rdesktop,vnc etc) to do practise labs.
If so, what was the server perfomance like, together with the specs (cpu,mem
etc).
Thanks in advance
spin
---------------------------------------------------------------------
Only those who dare to fail greatly can ever achieve greatly.
Blogs and organic groups at http://www.ccie.net
This archive was generated by hypermail 2.1.4 : Sun Mar 01 2009 - 09:43:39 ARST