I spent a decent amount of time today screwing around with VIRL in the cloud. The documentation is ok, and the youtube tutorial is good enough to follow along with if you have some experience. But I still don’t feel good about this “new feature” and I certainly wouldn’t recommend it to my students. Here’s why.
Reasons Why I’m Not Impressed With VIRL in the Cloud.
I have a few reasons why this new feature doesn’t impress me and here’s why:
You still have to have VIRL installed locally. What??? Really?? I cant just have a VM in the cloud that I instantiate when I need it? Well, from what I can gather with only about 4 hours of experience, you MUST have a local install of VIRL to even attempt to make this work. This is a limiting factor. I don’t understand why I can’t simply purchase VIRL and install it on bare-metal in the cloud without this requirement. Really though, I should be able to run this on Ravello, Amazon, or Google as I see fit. This is 2016.
After giving things a second go around I can admit that I was wrong here. There’s actually clear instructions on how to spin up VIRL on Packet without running a local VM. The process took only about 10 minutes to setup.
2. The Solution Requires OpenVPN connectivity
You have to install a third party VPN client. This is too much work to ask. I’m sorry. I shouldn’t have to spend more time getting this thing to work than I need to spend configuring etherchannel on a 3760 switch.
3. Because You Have To Type Commands From the Linux CLI.?
Why oh why do we have to type commands from the CLI? If I have to enter a single line from the linux CLI it’s a fail. People using VIRL to learn should be called upon to enter salt calls and terraform commands. The entire purpose of a paid product like this is to take away the overhead of building a lab. VIRL does NOT do that. It adds overhead by means of commands and verifications that are not necessary.
4. It takes 30 minutes EVERY time you build a VIRL cloud-based instance.
Uh.. really? Again, this is insane. I can spin up an Amazon server in less time. Why does this take so stinkin’ long to build? Oh yeah, because my local VIRL instance has to talk to my remote VIRL instance. Again, there is so much friction here that it makes me want to cry.
My Initial Stance
Initially I’m not at all impressed. The solution to run VIRL on bare-metal in the cloud is much to time consuming and way to involved to be worthwhile for anyone studying for Cisco certifications. A Cisco Learning Lab can be instantiated much faster and the money you spend is much more worthwhile in my opinion. I just cant believe that there is so much involved in getting this product working. Personally I believe that about 80% of the garbage hoops we are forced to jump through are because Cisco doesn’t want the product to be available to the public without charge. Frankly it makes me sick to think that I wasted over a hour of my day trying to kick then tires on a product that did not work when all was said and done. I could ping the server, but could never HTTPS to it and log into UWM. I couldn’t get VMMaestro to talk to the IP even though I had a VPN established. And I even had to install Viscosity as my VON client because installing OpenVPN straight away never worked.
While I can certainly figure out all the issues I encountered today and eventually get VIRL working on bare-metal in the cloud I’m still struggling to wrap my head around the benefit. Honestly, I would benefit more from an Amazon AMI that I can spin up in a few minutes and requires my key before I can login. This solution is too much of a cluge. I’m sorry. The VIRL team has made GREAT PROGRESS, but this would not be included in my list of things I’m thankful for. Please think about how you can offer this with ZERO TOUCH configuration in the cloud. Then I’ll praise the feature all day and all night. Maybe the folks at Ravello Systems can help make this a reality? Until then, I’ll call it quits and leave the lab frustrated.