Hi, I’m looking to buy a PC to run Cisco Modeling Labs. Can someone advise what is the minimum requirements I need to run CML? processor? ram?
Please advise.
Thank you.
Hi, I’m looking to buy a PC to run Cisco Modeling Labs. Can someone advise what is the minimum requirements I need to run CML? processor? ram?
Please advise.
Thank you.
Hello Kader
You can find the official requirements for CML at this link here:
However, the answer is not quite that simple. To run CML on a Windows PC, you need to plan for a Linux-based CML controller VM plus all the virtual network nodes you’ll run in your labs. While there are “official minimum” specifications in the link I sent, real-world experience shows you need significantly more resources for practical study.
The primary bottleneck for CML is RAM. Every router/switch/node consumes dedicated RAM, and running out of RAM causes nodes to crash or to fail to start. The following is a guideline for the amount of RAM you’ll need for what you want to achieve:
As you can see, the IOSv and IOSvL2 nodes take up just a bit of RAM (512 MB to 1 GB) while CSR1000v (3-4GB), IOS-XR (4-8GB), and NX-OS(5-8GB) require progressively larger amounts of RAM each!
Other than RAM, it is recommended to use INTEL CPUs, because Cisco states that “CML is only fully supported using Intel CPUs…Support for running CML on non-Intel processors will be provided on a best effort basis.”
Finally, here are some recommended setups that are suitable for differing degrees of requirements:
Budget Build (CCNA study):
Recommended Build (CCNP/CCIE Enterprise): ![]()
Power Build (Data Center/Service Provider, large labs):
If you are serious about networking and are planning on studying for the long haul, this is an investment, and it’s worth spending that little bit more to ensure a longer life to the system you’ll be purchasing. I hope this gives you a good guideline for your purchase. When you do end up choosing a system, let us know what you got and how it’s working out for you!
I hope this has been helpful!
Laz
Hello.
I’ll also share my own experience since I decided to be bold and ran CML with only 10 GBs to study for both ENCOR and ENARSI. Consider these as “minimum” requirements ![]()
Suprisingly, 10 GBs of RAM was actually enough.. You will typically run around 2-12 nodes MAX, I never had to run more than 10 myself except for some particular cases such as BGP or multicast routing.
Do not run the IOSL nodes, they do not work properly in 99% of the time. The IOSv nodes each eat around 512 MBs of RAM and are perfectly fine to run.
However, for certain features, you will have to run a CSR router (IOS-XE) which eats up to 3 GBs of RAM, so you could only run 3 of them ![]()
Either way, I agree with what Lazarus said. Stick to the recommended build since running it on requirements that were as low as mine is a risk after all ![]()
For CCNP Service Provider, I had to upgrade from 10 GBs to 128 GBs considering that SP runs IOS-XR nodes which are as large as a gaming PC.
David
Hi Laz & David,
Thank you for the input. I will check out the site given.
Best,
Kader
Hi Kader,
One thing I can add is that usually the bottleneck is RAM, not the CPU. When your devices boot you’ll see a spike in CPU usage but once the dust settles, it’s ok.
Mini PCs are an excellent option. An Intel NUC, Lenovo ThinkCentre, Dell Optiplex, etc.
These draw not much power when they run idle. You don’t need the latest generation CPUs, an Intel i5 10th or 11th generation works great. Add 64gb of RAM and you’ll be able to run almost everything.
Rene
Hi Rene,
Thanks. I was looking at a Dell Optiplex but the price range was similar to that of buying a laptop. Even the refurbished ones were a bit on the high side.
I will be looking at the new models that have DDR5 and have the capacity to upgrade to 96gb ram.
Thank you
Hi Kader,
The new ones are all very expensive but an older refurbished one is usually not too bad. DDR5 is very pricey at the moment.
On an older unit you could install proxmox and then CML in a VM. Idle, those micro PCs consume 10-15 watt or so.
Good luck ![]()
Rene
Hi Rene,
Thanks for the info. I will look into it.