Since the early days of Big Data, DiffusionData’s Diffusion intelligent data platform has delivered a powerful solution for real-time analytics and data integration. Across the landscape of rapidly digitizing businesses – ranging from the Internet of Things (IoT), telecom, edge computing, healthcare, energy, end-user messaging and more – the Diffusion platform simplifies the challenge of delivering data from diverse sources to the end users who need it. Its proprietary algorithms combine nimble streaming analytics with easy-to-build, flexible SDKs for the full range of client use cases, supporting enterprise and startups alike.
The Challenge
Upon transitioning the Diffusion Cloud platform to a microservices architecture, DiffusionData experienced a rapidly broadening influx of feature requests and integration opportunities. Dockerizing the application infrastructure made “last mile” cloud-based service scalability easier to manage, crucial to the SaaS business opportunity.
However, these improvements exposed new bottlenecks in back-end data infrastructure dependencies, particularly in effectively maintaining security policies and existing customer hybrid datacenter integrations. Customers running portions of the Diffusion platform inside the datacenter were not able to leverage the latest DiffusionData product innovations. The Diffusion Cloud product development team faced an accelerating backlog that jeopardized crucial growth opportunities.
The Solution
As an AWS SaaS Competency partner, CloudGeometry brought a wealth of experience working with startups to uplevel the software development pipeline in pursuit of more efficient release integration and deployment. Adopting the best practices of Infrastructure as Code (IaC), the CloudGeometry team built out complete, modularized provisioning logic, such as in certain cases moving to Helm from Terraform. Adding a single managed service for using Vault relieved application developers on the product team of the burden of redundant roles and policies management originally built into separate back-end data services. By automating all configuration and validation processes on one GitOps pipeline, the DevOps experts at CloudGeometry simplified release of new features to legacy data-center customers who had not been able to enjoy the latest/greatest Diffusion Cloud features.
The Benefits
Because the Diffusion Cloud infrastructure already ran on AWS EKS, it could quickly leverage the new CI/CD chain as a single source of truth for deployment logic. It eliminated configuration errors between Dev, QA, Pre-prod, and prediction environments. Quickly put into use, it allowed customers to try new features sooner. Moreover, the product development team could confidently add features to both backend and frontend without having to take the time to debug dependencies previously exposed only at the end of the release process.
Key Features
<div class="case__txt--cols"><div><h4>Secure, Automated Key Management</h4><p>Hashicorp Open Source Vault calls AWS KMS, completely automates secrets encryption and authentication processes requiring no human intervention.</p></div><div><h4>Better granular provisioning of EKS running on EC2</h4><p>With application K8s pods provisioned via Helm distinct from infra Terraform modules, developers can more easily enable scaling and application state control.</p></div><div><h4>Maintain Docker compatibility with on-prem stacks</h4><p>A single Infra as Code provisioning chain makes new application features compatible even with non-AWS environments for legacy customers.</p></div></div>