I am working my way through some lab examples in an eve-ng lab. I previously was using vSRX-NG images, but they are just too heavy, with all that CPU reaming polling going on - so I tried vRR instead. Here are some of my obervations so far, I would love to hear from other people using them for this purpose.
- When they work, they work super well. Nice and snappy!
- These things are lightweight. They run with a single vCPU and 2GB RAM. As they don't use DPDK to emulate a real FPC, they don't smash your host CPUs
- They seem to support all the routing protocols I need to complete the book, however I haven't really gotten into the tricky stuff yet due to some of the...
- They still take a while to boot, something in the order of 11 minutes from cold start to pingable management interface
- The interfaces are named emX (0, 1, 2 etc) which might catch some people off guard. No ge/xe etc. No FPC!
- They are a bit unstable. I'm using 3 variants (19.2R3.5, 20.4R3.8, 23.1R1.8) and all 3 occasionally fail to boot. Sometimes they spontaneously end up in a core dumped state, unable to read their own config. Other times, two directly connected boxes which were working can't ping each other (wireshark of the link between them shows nothing but unanswered ARP)...
The cons are annoying, as the vRR does seem like the holy grail for virtual lab use.. I just can't quite get them to behave reliably. I haven't messed with any custom templates or options, but maybe that will be necessary.
Overall so far when they are in a working state, they're neat - and definitely seem to be the way to go for casual labbers not wanting to use up too many resources to test out a few protocols.