Last week, Delta Air Lines faced a technical hitch. On the surface it wasn’t an enormous problem - a power outage at their Atlanta data centre caused a switchgear to fail (like a circuit breaker in your home). However, the backup systems didn’t come online correctly and the failure of this one piece of equipment then caused a complete shutdown of the Delta IT systems globally.
Over 1,000 flights were cancelled on Monday and over 500 on Tuesday. Passengers are being refunded in full and given $200 vouchers as compensation. All passengers stranded in airports had to be provided with hotel accommodation and food. The cost of the system failure isn’t known yet, but a similar incident a month ago at SouthWest cost the airline tens of millions of dollars to resolve.
The New York Times doesn’t report whether the Atlanta data centre was an in-house operation or outsourced to an IT supplier; however a quick dig around on various travel- and IT-related blogs shows that many commentators are pointing a finger at outsourcing being to blame for the Delta catastrophe.
In this specific case, I’m not aware if the data centre is outsourced or not, but it strikes me as odd that online commentators would immediately blame outsourcing in this way. It’s clear how vital the data centre is to the safety of the fleet and for the smooth operation of the business - just an outage of a few hours can cost millions in lost business, compensation, and the vanishing goodwill of customers who will never choose to fly with that airline ever again.
If this part of a business is clearly so critical then surely it makes sense to hire experts to manage it. Would you really locate a critical data centre in your own office facilities and hope that you have the right team available to keep it running 24/7 or go to a supplier with global facilities and the best team of experts in the world?
The answer seems obvious. Outsourcing is about ensuring that critical facilities are up and running much more than reducing costs. The critics who say that a data centre failure must have been caused by outsourcing clearly have never had to work in an environment where system availability is critical. Outsourcing this critical infrastructure is probably more expensive than managing it in-house, but the cost of one single error is enormous. It makes sense to reduce the risk of failure by building storage or processing power in the cloud with a partner that can guarantee their system will never fail.
Some people will always criticise outsourced projects and services. Outsourcing doesn’t always work. The media has documented many failed projects, but when planning critical back-office infrastructure that simply cannot fail outsourcing to a world-leading expert seems like a sensible strategy and not an approach that would increase the risk of failure.