Have a Subnetting Convention question

So I've been studying subnetting extensively for the past few days, trying to brush up on what I learned in college. What I am most confused (and can't seemingly find an answer to) is common convention and the "right" way to subnet an internal network.

I just joined a company and looked at their Dell Sonicwall. The IP addressing scheme we have is 192.168.21.X /23. From what I understand, the /23 indicates that the network address is 192.168.20.0 and the broadcast address is 192.168.21.255.

So I have a couple questions:

1. How many subnets are actually being made here? I understand that 512 (addresses made) and 510 (usable hosts) are being made, but is this all under one subnet? How does this /23 work in a Class C subnet? I think it's only one subnet being made but I'm not sure. I'm just really getting confused with the whole "borrowing bits" thing that is relevant in Sub-Class C subnetting and not in Class C addresses with non-sub-class C subnetting.
2. If I made a separate network 192.168.1.0, would this be able to communicate to 192.168.21.X? My feeling is yes, but will need a router to communicate.
3. An example I saw online said that if I had a subnet of /23 or /22, that if my network address was 192.168.1.0, that I would have 192.168.1.0-192.168.3.255 and 192.168.1.0-192.168.7.255 respectively. If I use a different 3rd octect, what happens there? Does the same principle still apply?

EDIT: So I was getting myself confused over the different formulas and rules and outdated Class-C configurations that I realized it's exceptionally easier to just do it all in binary. Like, much easier. Thanks for all your comments.