| In the 1980s the computer industry experienced a massive transformation when core software system architectures changed from being mainframe-based to client-server. This shift changed virtually everything in software from the hardware, to software designs, to the practices for development and operation of that software. Today we are in the midst of the next radical change as the industry moves from client-server to cloud-native. Sometimes called the third platform, cloud-native is characterized by highly dynamic systems, where every element, from the hardware and operating system, to networks and software deployments, are in constant flux. Whereas for second platform systems (client-server), software was written with an expectation that the systems that it executed on were quite stable, in the new world software must be written, deployed, and managed in a manner that anticipates change. That is, the software that runs on a highly distributed, constantly changing infrastructure must have zero downtime even while the lower layers are shifting about. In fact, applications must have zero downtime even while the application itself is changing, due either to an upgrade being performed, or to the application itself experiencing trouble (there’s a bug!). Just as the shift from first platform (mainframe) to second (client-server) changed everything about the way that software is constructed and managed, so does the shift to third platform. Software practitioners must learn a whole new set of design patterns as well as master new software engineering and management tools and methodologies to remain effective. Ultimately our aim is to provide reliable digital solutions even while the infrastructure they are running on is unstable. It’s not Amazon’s Fault On Sunday, September 20, 2015 Amazon Web Service (AWS) experienced a significant outage. With an increasing number of companies running mission critical workloads, even their core customer facing services on AWS, such an outage can subsequently result in far reaching system outages. In this instance, Netflix, Airbnb, Nest, IMDb, and more all experienced down time, impacting their customers and ultimately their business’s bottom lines. The core outage lasted more than 5 hours (or even beyond, depending on how you count), with even longer AWS customer downtimes before they had their systems fully functional. |