Dynamips - not bad

kingpinofdisks

I had a customer service 'incident' this morning with the place I go for rack rentals.

Now I am going to use a differnet rack rental place.

This also caused me to take a first look at Dynamips.

Not bad. I need to acquire a couple of more IOS images for it.

My host OS is Win2k3 Ent Server X64, running on a Athlon X2 4200+ with 4gb RAM.

I used GNS to create my lab file, then cleaned it up for use with Dynagen, which seems to use less resources.

I am lazy, so I just made a screen-shot of the layout in GNS and printed it for reference while I am lab'ing it up.

If it keeps running well, the only lab time I will need online is for Switching, and some of the ONT labs.

I am currently running a 5-router OSPF lab with 3 frame & 2 ATM links - it seems to be using about 1.2gb of RAM and my CPU hovers around 25% when the routers are just sitting there, with spikes up to 100% on an irregular basis.

This box is also running 2 VMWare images (1 XP and 1 2k8 Server) at the moment, so I am impressed that I can connect them all up.

_maurice

Are you using GNS3?

cisco_trooper

You shouldn't be seeing more than a 2 or 3% CPU utilization with that server and only 5 routers running. Check out the idlepc settings in dynamips. Last lab I did in dynamips was 14 routers and it ran right around 2% CPU on a dual core 2.8 with 4GB of ram. Host OS is Server 2K3.

Maybe the VMs had something to do with it, but it's work a look if you are going to start creating larger topologies.

cisco_trooper

Also, GNS3 is a resource hog...I personally just stay away from it. It isn't THAT hard to set up your dynamips config once you've taken the time to learn how to do it.

Mrock4

I second the notion that GNS3 seems to use a ton more resources. I type up a NET file or just use a template, and modify it as necessary to get up and running. Also, I don't know if it was me or not, but when I used GNS3, it'd take a couple of attempts at setting the idlepc to get it right. I can generally get it right on the first go with Dynagen.

kingpinofdisks

cisco_trooper wrote:

You shouldn't be seeing more than a 2 or 3% CPU utilization with that server and only 5 routers running. Check out the idlepc settings in dynamips. Last lab I did in dynamips was 14 routers and it ran right around 2% CPU on a dual core 2.8 with 4GB of ram. Host OS is Server 2K3.

Maybe the VMs had something to do with it, but it's work a look if you are going to start creating larger topologies.

Maybe so if Dynagen were the only thing running. I have 3 VMs running, which is why all the extra CPU. This desktop is my virtual lab for everything now.

The IdlePC settings are fine AFAICT.

I setup a lab last night consisting of 3 2k8 servers (LA, Chi, NY), 3 3640 routers in a frame cloud, and 1 Trixbox VM (for simulating voice). Kinda cool

keenon

kingpinofdisks wrote:

cisco_trooper wrote:

You shouldn't be seeing more than a 2 or 3% CPU utilization with that server and only 5 routers running. Check out the idlepc settings in dynamips. Last lab I did in dynamips was 14 routers and it ran right around 2% CPU on a dual core 2.8 with 4GB of ram. Host OS is Server 2K3.

Maybe the VMs had something to do with it, but it's work a look if you are going to start creating larger topologies.

Maybe so if Dynagen were the only thing running. I have 3 VMs running, which is why all the extra CPU. This desktop is my virtual lab for everything now.

The IdlePC settings are fine AFAICT.

I setup a lab last night consisting of 3 2k8 servers (LA, Chi, NY), 3 3640 routers in a frame cloud, and 1 Trixbox VM (for simulating voice). Kinda cool

I would mention that the image, memory specified and cards installed will make the idlepc vary. I have changed one router from 64 to 96mb ram with all the same cards and the utilization changed greatly.