I would like to simulate or emulate the design of a replacement network which will be L3 routed (the easy part) but will have a lot of simultaneous L2 VLAN bridging (or tunnels) to accommodate the transition from the existing network.
I am perfectly happy to get the VIRL subscription from Cisco, my question is not about cost, but I am unable to tell if it solves my problem, either alone or with GNS3, and would prefer to avoid wasting the time if the answer is "not a good approach".
The plan is to use EIGRP and L3 routing mostly, but during the transition have to bridge a lot of legacy VLAN's (and/or tunnel them); we also want to experiment with various configurations of two microwave rings involved (easily simulated by an L2 switch in various configurations) and how those interact with the p2p fiber connections used for a second or third redundancy. So the closer to functional equivalency the better. All private fiber or microwave, no VPN or encryption involved at this point, no ASA's (we will have some but that's a separate and easily handled aspect) and no IOS routers just the L3 switches.
I have GNS3 running (never used it before) and it is very cool for old IOS routers, but I am reading conflicting info (perhaps due to changes over time) on how well either VIRL alone or GNS3 with IOSVL2 and IOSVL3 images from VIRL can emulate catalyst switch functionality (especially L3). I have yet to find any side by side feature comparison of what will, and will not, be similar to the Catalysts. We will be using 3650's and 3560CX's at most locations (about 15 total so under the VIRL 20 limit for PE).
So basic question: How close can I get to a reliable simulation in this way?
Related question: Assuming I can find the functionality I need, how reliable is it to assume if it works in simulation it works in reality the same way (again, performance aside)?
Bonus question: how close will the configs be to actual configs needed for the Catalysts?