GNS3 – Conducting Latency Tests

I'll go ahead and provide some more "in-depth" instructions on telling GNS3 to use a "remote hypervisor" (a dynamips instance listening on another machine).

You can even use this setup with Amazon EC2 (or another utility computing provider). A setup with a utility computing provider would also be extremely useful for testing out topologies with a very large number of routers, to see how OSPF may behave at scale, etc. I've documented how to do this at my last job (but this was before GNS3 was so popular) and we used it successfully for numerous scale tests. This would also be ideal for folks that don't want to spend the money on a high-end server but can afford to spend a few dollars for a few hours of "server time" for GNS3 labs, without chewing up CPU cycles on their home machines.

The following instructions assume your remote hypervisor is running Linux.

Step One: Set up your Linux box.

jjensen@jjensen-dev:~$ mkdir dynamips
jjensen@jjensen-dev:~$ cd dynamips
jjensen@jjensen-dev:~/dynamips$ mkdir tmp

Step Two: Set up GNS3. Edit -> IOS images and hypervisors. Click the "External Hypervisors" tab. Add in the IP address info for your remote hypervisor (Most of the time you can leave all the fields besides IP address as the default). Once you've filled in the IP of your remote hypervisor, click "Save".

Once you have your remote hypervisor added, go back to the "IOS images" tab and add in IOS images. There are two important bits to remember:

Use the "Linux format" directory notation (see screenshot), and make sure that the "use the hypervisor manager" checkbox is unchecked so that you can select the remote hypervisor that you just added. When adding the IOS image with the remote hypervisor, make sure that your remote hypervisor is selected.

Once you have your IOS image added, start up Dynamips on your remote machine.

jjensen@jjensen-dev:~/network_eng/dynamips$ sudo ./dynamips-0.2.8-RC3-community-x86_64.bin -H 7200 &

You should now be able to add devices to GNS3 topologies on your Windows machine, and GNS3 will communicate with the remote hypervisor on another machine. You also get the console access via PuTTy and such for free as well. Note that there isn't really a "configuration limit" on how you want to add your IOS images and hypervisors. You could even have it so that you have a separate remote hypervisor per IOS image if you wanted. Pretty powerful stuff.

If anyone could add additional answers on how to make this work with JunOS/"virtual" Olive, I think that would also be pretty useful. :-)

Bell is telling you the truth. When you try to push 5Mbps (or more) into a 5Mbps connection, everything files into a neat little order (read: queue.) Your ping goes out without delay because there's no backlog. The reply, however, is now at the end of the queue. TCP is doing exactly what it's supposed to here -- the sender is filling the allowed receive window.

There are things you can do on your side of the line (QoS, WRED, etc.) to help reduce the effects, but this is the sort of thing you're going to see when there's a large difference between sender and receiver bandwidth. I've lived with it for years (T1, 6Mbps DS3, even 10Mbps cablemodem) You could ask the ISP to reduce the queue size on their side, but they're not likely to do it, as it'll result in packet drops.