McDonald’s was hit by a system failure Friday that closed restaurants and disrupted online and app orders around the world, including in the United States, Australia, Japan, Hong Kong, and the United Kingdom.
My question is why are the systems designed to be dependent to upstream services 24/7? Wouldn’t a better approach be to have systems that can run disconnected, then simply upload/replicate data when a connection returns?
These are franchises, right?
I deployed such POS (Point of Sale) systems in the late 90’s, because connectivity wasn’t ubiquitous then. They were designed so franchises could upload/replicate however you needed: continuously, when a connection was available, on a schedule, etc. Some places had pooled telephone lines to achieve the needed throughput.
I get the mobile ordering being impacted, but why would you tie the local kiosk to a web service?
There are many considerations where the fault lines should be. Somethings absolutely need upstream support and cannot just be buffered… for example payment processing.
Shit happens and we DevOps people do everything we can to minimize it. That’s why your apps might go down for like 2hrs a year.
Add comment