Making The Best Use Of Public Cloud Infrastructures

Waylan Johnson, Vp, Cloud Architecture & Operations, Swbc

As a diversified financial services company, SWBC is always looking for a competitive edge with technology for the business world while maintaining the highest standards of security, compliance, and availability. In days long past, we would build a data center full of physical servers, storage, and network capacity at a pre-planned high-water mark. Virtual servers on converged and hyper-converged infrastructure ushered in a way to gain economies of scale; however, capital outlays and time to delivery were still an issue. The arrival of public cloud allows us to deliver solutions with increased resiliency to our clients at a faster speed that does not compromise our low tolerance for risk.

Our company has had an internally developed payments application for both financial institutions and individuals alike in the public cloud space for several years now. We leverage non-relational database services where the cost is based on throughput, not storage, and scales automatically. Highly available, distributed, near real-time search capabilities are paired with this database in conjunction with serverless compute, for an extremely resilient low-cost, transactional-based application that we only pay for as needed, one function call at a time. This application can be hosted across multiple regions both for resiliency and delivering content closer to the request; therefore, speeding up the application response time since the speed of light is constant.

The technologies and managed services used by our payments application in the cloud serves as a blueprint and planned end state for many of our current and future applications, as we move through the transformation. We can leverage the elasticity of cloud resources across different availability zones or regions in ways we could never achieve running our own data centers or even utilizing colocation facilities. Providers like Amazon Web Services (AWS), Azure, and Google Cloud Platform provide all the infrastructure and the API’s to create the resources we need where we need them through software-defined means.

Our plan is optimizing workloads, attempting to break monolithic applications down so we use alternative licensing models for databases, moving compute to immutable containers or serverless architecture where we can, and introducing auto-scaling capabilities. For example, in the past our tax-related application servers had to be built for that high-water mark during tax season, but went underutilized the remainder of the year. Without either third-party applications that perform supply and demand management or manual intervention, this would turn into wasted compute cycles. Now in AWS, our Cloud Operations team can leverage Cloud Watch to monitor these resources to create alerts, Lambda calls to act upon those alerts, and Route 53 with Elastic Load Balancers to shift from scaling vertically like we did on premise to scale horizontally in the cloud. These types of transformations cost only a few dollars per month instead of the thousands per month in third-party applications or labor.

Moving forward in collaboration with our AppDev teams, we will be focusing on blue-green deployments with infrastructure-as-code. Using Terraform (which is agnostic to specific Cloud providers) scripts together with Packer, we will automate building out new Center for Internet Security benchmarked server images with a fresh set of patches in parallel with the application release pipelines. If a release passes all user acceptance and regression tests, we destroy the old servers and promote the new ones. All this is achieved simply with code, resulting in our ability to accomplish more with less. The ability to define and provision a datacenter infrastructure, to create and destroy within minutes, is one of main reasons an organization chooses to migrate to the cloud. Subsequently, we expect a significant reduction in secondary storage since backups will be limited to our code repositories rather than entire application servers. We also anticipate being able to redeploy resources and efforts such as monthly patching into other areas. Furthermore, our Cloud Innovation team has recently starting using machine learning capabilities within multiple cloud-based services to create robotic process automation and chatbots to assist our internal customers with repetitive tasks, such as self-service and application functional testing for monitoring availability.

All these efforts could not be achieved without public cloud offerings to help build platforms that are secure, resilient, self-healing, scalable yet help maintain lower technology costs and allow us to explore our creativity and innovative capabilities with a higher velocity of delivery to our clients and customers. Our mission has always been to pioneer financial and insurance solutions through a winning combination of world-class service and revolutionary technology. The transformation landscape of public cloud will ensure we continue to achieve our mission, vision, and values.