Network fundamentalism is not a bad thing

By Chuck Yoke
Network World, 07/02/01

During a recent period of severance, I interviewed for a variety of positions, one of which was for a network architect/engineer. Being more of a manager than an engineer, I didn't expect much out of the interview, but I went anyway.

The interview entailed standing in front of a group of network engineers and discussing a variety of networking topics. I was asked to explain how a workstation talks to another workstation on the same subnet and on a different subnet. I was quizzed on routing protocols, routed protocols, Open Shortest Path First (OSPF), Routing Information Protocol and Border Gateway Protocol (BGP). I was asked to explain how distance-vector and link-state protocols worked. I was grilled on switching, routing and Recursive DNS. Because I was not actively involved in many of the areas that were discussed, I left feeling that I was probably not the caliber of employee they were seeking.

Imagine my surprise when I received an offer a few days later. It appears that I had been only one of two people interviewed who understood Proxy Address Resolution Protocol (ARP). I was also only one of two people who could explain how an IP address is resolved to the media access control address, how a router handles the Layer 2 communication process between two workstations on different subnets, and what the difference is between link-state and distance-vector routing protocols.

Even though I got tongue-tied trying to explain Recursive DNS, fumbled my way through BGP and completely blanked out on OSPF stub networks, I was still viewed as one of the top two applicants because of my ability to describe the very basics of IP networking.

While I decided not to accept their offer, the whole process made me stop and think. Is this the true state of network engineering today? Have employers placed so much emphasis on the "process" of networking - creating network diagrams and configuring routers - that the "foundation" of networking - how it all works - is being neglected? Has the industry become so fixated on certifications that we are being deluged with two-week wonders who can configure a router but have no idea how a routing protocol works? Have we created a workforce that is analogous to a mechanic who can change the spark plugs but has no idea how all the parts fit together?

Some people may believe it's not important for network architects and engineers to understand the underlying protocols. But in my mind, unless you understand OSPF, you can't configure it on the router. Unless you understand ARP, you can't troubleshoot connectivity problems.

I wouldn't trust my car to a mechanic who had no idea how the distributor, spark plugs, carburetor and crankshaft work together to make the wheels turn. And I wouldn't trust my network to an engineer who has no idea how ARP, Proxy ARP, DNS, network address translation, Simple Mail Transfer Protocol/Internet Message Access Protocol, routing protocols and routed protocols all work together to get this column from my hard drive to my editor's in-box.