The cloud is without any doubt the new model for enterprise IT. Tomorrow a large part – if not the whole – of enterprise IT will be in the cloud (but beware of easy shortcuts… cloud does not always means Saas).
Cloud taken in the broad sense and Saas in particular have undisputable advantages for businesses : frequent updates, lower deployment and maintenance costs. A rational choice.
But it’s not a panacea. Saas forces to a certain form of standardization that is does not always aligne with the will to use IT as a differenciating factor. Hence the fact some large businesses come back to “custom software” for some strategic and differenciating projects.
But doing custom software does not means one should avoid the “cloud rules” (or use cloud in a Paas or Iaas model) and Delta Airlines gave us the perfect example this summer.
For those who did not follow the news : Delta experienced a major IT outage this summer. A power outage in Atlanta (where is the company’s HQ) lead to an IT outage and the cancelling of thousands of flights.
While most major airlines decided to use softwares from the market, mostly in cloud, for their passenger Service System, Delta decided to develop its own one, Deltamatic, from scratch. The result was significant benefits : not only Delta excels in operations and flight management but the airline is also able to launch offres, bring changes to its flights program or loyalty program in a very short amount of time because it has lots of in-house developers ready to improve the system when necessary. For Delta, not chosing one of the majors cloud/saas players of the market had a huge payback…until this summer.
Such an outage would not have had the same consequences for another airline because of the very nature of cloud infrastructure with backups and redundancy that make that even if the main infrastructure has an issue, other servers can take over. In theory, in-house or not, one could have excepted Delta to design its infrastructure the same way and many people were surprised by this summer’s incident.
Delta’s mistake was not to do everything in house. Even if it may look opposite to what the rest of the market is doing it may make sense. But the the mistake was to bet on a on premise model instead of industrializing its own software in the cloud.
That’s an important lesson for whom is thinking about doing custom software to be differenciating. It may pay when you have both the vision and the capabilities to execute. But that’s not because you don’t choose solutions from the market that you may not follow the same rules to design your infrastructure.
You’ve been warned…