Macquarie Bank sees its move to Google Cloud as a “second opportunity” to experiment with new automation infrastructure provisioning and management approaches. At the All Day DevOps conference, Jason O’Connell, CTO for banking and financial services, stated that the bank manages its Google cloud environment using Kubernetes.
“Last year at Macquarie, we launched our second public cloud, Google cloud,” O’Connell explained. “This offered us the chance to consider an alternative approach to cloud management, reconsidering what we’re doing with the cloud.” “Because it was our second chance, our second cloud, we had the opportunity to do things differently.” “We’ve been running Kubernetes in production since 2017, so we’re quite early Kubernetes users for a number of major workloads surrounding our digital banking offering, and as a result, the team was highly skilled and experienced at operating Kubernetes itself and running workloads there.” “So, when we heard about a new way of managing the cloud in that we could use Kubernetes – and Kubernetes Operators – to handle the cloud itself and do other automation, we assumed it would be an incredibly different point of view.”
This operating paradigm was previously revealed by Macquarie at a Google cloud conference in September, while the All Day DevOps presentation offered far more insight on the endeavor. According to one description, operators are pieces of code that “function as an extension of [an] technical team, monitoring over a Kubernetes environment utilising its present state to make real-time choices.” According to O’Connell, the notion of deploying Kubernetes Operators to govern cloud environments has risen in popularity in recent years. “I believe you’ll hear a lot more about this type of concept in the future,” he said. “You’ll note that AWS now has ACK which lets you to accomplish the same thing that I was talking about with Google, and Crossplane.io is another provider that also gives cloud management via operators.”
Macquarie employs operators to provide the resources required to onboard an application or workload into Google Cloud. The operators lie between Git, which describes the “desired state” of infrastructure required for a program or workload in code, and Google Cloud, which has the “target state” or actual state. They are continually polling for ‘events’ that would signal a change on either side and then seek to reconcile the change such that it is reflected in both the intended and target states.
