0

I'm running some performance tests on GNS3 (I know I shouldn't, but it is not until Friday that I will have the college lab available), and I'm trying to compare real BW or "speed" and theoritecal.

So if I just put the command : "show interfaces serial1/1"

Do I get the "theoretical" value of the link?

(MTU 1500 bytes, BW 1544 Kbit, DLY 20000 usec)

So when I run iperf from my loopback interface (PC1) (client side) to my Linux VM (server side) I get and average of BW= 1.1 Mbps.

Is is safe to say that the performance of the link is:

Performance = real BW / theoritecal BW = 71.24 %

This is the topology I use, and it goes from PC1 to PC2:

enter image description here

I expect the performance to be higher in a real environment since GNS3 emulates everything and it consumes too much.

This is my real PC specs: Intel i5 processor 8 GB of RAM

Thanks in advance! :)

Daniel

5

Don't trust GNS3 for throughput tests. As soon as you use a router in your topology you are likely only to get 5mb/sec tops, regardless of whether you use a serial, ethernet or fast ethernet interface and regardless of how powerful your PC is. (Note that the performance hit won't be seen if you use a GNS3 switch instead of a router, however, the switches are extremely limited in what they can do).

A few years ago GNS3's website used to say something along the lines of don't use GNS3 for production workloads as GNS3 routers perform more than 1,000 times slower than the real thing. I know you're only labbing, but the reason why I mention this is that the devs themselves say its performance is not even close to what you will get from real equipment.

Finally, just an FYI - A little while back I did a few iperf tests through a couple of VMs connected to GNS3 on my i7 with 32GB of RAM (see this page for specs, though at the time of writing I only had 16GB of RAM). If I recall correctly I only got about 3mb/sec. However, when I rejigged the lab to include a couple of Riverbed Steelheads I saw a massive improvements, but that's a whole other story :)

To sum up the above - GNS3 is great for testing protocols, configs, etc but is pretty much useless when testing throughput.

| improve this answer | |
  • No worries at all, it's my pleasure. – OzNetNerd Nov 30 '15 at 21:41
  • 2
    GNS3 developer here. I confirm the answer we can't get reliable performances it's depend of a lot of factors. The images from constructor are sometimes rate limited in virtualized environments for example. – Julien Duponchelle Dec 8 '15 at 10:28

Not the answer you're looking for? Browse other questions tagged or ask your own question.