Redundant WAN on the cheap

So, I’ve been trying to find the best way to provide some extra Internet bandwidth at work without breaking the bank. My initial thought was to double up my T, until I realized how much that would cost me per month, and I still wouldn’t come close to the speed of my home cable modem connection. So, I’ve decided to bring in a Comcast business cable modem as our primary “web” connection. We’ll maintain the T for published services, outgoing email, and redundancy. Simple, cheap, great.

With the easy part out of the way, I embarked on a quest to manage two WAN links. Our firewall/gateway is a Microsoft ISA Server, which doesn’t support multiple WAN links. The only ISA add-on that does support multiple WAN links has just been deemed end of life by EMC. Just as well, as it was $3000. So, I began looking for hardware solutions. Thus began the hard part.

I found many SOHO solutions that can load balance multiple WAN connections. The list includes the NetGear FVX538, Xincom DPG602, Linksys RV082, and a few others. All of these are low-cost, feature rich solutions, but are obviously built on low-cost hardware. They don’t seem capable of handling the throughput of a large network, as evidenced by their 253 user limit. I’m not even certain what this user limit is, or how it’s enforced. In any event, these seem best suited to small offices of 50 PCs or less. Certainly not ideal for our 300+ node environment.

My next foray led me to the Cisco 1841 router. There are a few users on the ISA Server message boards that seem to be raving about this router. It has many features and options, but my initial impression is that it cannot handle the routing speeds of a broadband cable connection. My CDWG rep is looking into this, so I should have some more information shortly. If it can meet my needs without breaking the bank, I think it’s a sure bet. Enterprise class features and support are always a plus.

Finally, an intriguing option would be to setup a Linux-based router using this article as a guide. This is option has the appeal of low cost, many possibilities with traffic control and QOS, and of course all the smell of a fun project. However, support and service are realistically nil, so the long term outlook isn’t good.

Hopefully I’ll have some more options and answers in the coming days, and I’ll post an update then.