Many of the details of how to set things up here are specific to your Watchguard box. I've never worked with those devices, so I'll speak in generalities. I'll also assume that you've figured out how to configure a site-to-site VPN tunnel between the Watchguards (which I'll refer to as "routers" from here on out).
- A site-to-site VPN tunnel needs to be established between routers
- A static route needs to be set on Site A's router to route traffic destined to 192.168.1.0/24 through the VPN to site B
- A static route needs to be set on Site B'S router to route traffic destined to 192.168.0.0/24 through the VPN to site A
- Firewall rules between sites should be set up using a "default deny" policy, which requires you to explicitly specify Access Lists for what IP addresses and ports are open between sites. For instance, if Site A users only need access to a file server and a printer in Site B, firewall rules should be put in place that only permit that traffic, and no other traffic. This limits collateral damage if, perchance, a workstation gets infected with a virus, it would not be able to spread through the VPN to systems in the other site.
- You probably want some sort of monitoring mechanism to alert you if the tunnel drops
Steps 2 and 3 will likely be set up automatically by the router when you configure the VPN tunnel.
Be aware that intra-site network performance will likely suffer quite a bit after this change. It's true, the available bandwidth of the cable connections is higher than the T1. However, your latency will likely increase by an order of magnitude (which is a huge deal for protocols like CIFS, which are very chatty and suffer greatly over high-latcency connections), and the reliability of the WAN will go down significantly. Even if these are "business class" cable internet connections, they're of a different class of service than a T1 is.