Data Center: Too Much High Availability Bandwidth?

One of my first engineering professors was pretty old school, having earned his reputation long before the days of computer aided design, finite element analysis and precision engineering.  He was the first one to teach me about the ‘safety factor’ – something that every engineer already knows intuitively: “Take whatever you need to support the load, and triple it”.  If you were tasked with building a bridge that needed to support 200 Tons – build it to support 600 tons!

Over the past few days, I have recently been accused of ‘over-engineering’ another part of the Data Center – our Internet Backbone Providers.  While some Data Centers are comfortable having 2 connections to the backbone of the Internet, the lifeblood of our business – we added our fourth connection this week.  Yes, that is four separate Gigabit connections to four different bandwidth providers, any two of which can support the full capacity of our Data Center.

For 14 years, we have taken the uptime of our data centers and core network very seriously, learning along the way of course, though sparing no expense or effort to offer our clients the best assurance possible that their Data Center would be up to the task at hand.  We have always purchased twice the bandwidth needed by our clients, and will continue to over-engineer and reinvest in the data center to give our clients the piece of mind they need.

I wonder if that professor of 20 years ago had any idea the simple lesson taught would apply in many more ways than just designing mechanical things?