Internet Scale Content Management with Apache Oak on Kubernetes
Content is a key enabler for many HP devices, be it multi-function printers, high speed commercial printers, 3D printers, or immersive computing platforms for creative solutions like HP Sprout. HP is in the business of transforming content from digital to physical and from physical to digital. Enabling this transformation keeps our devices constantly connected to cloud services.
HP runs a global scale federated content management system implemented with Apache Oak at is core. Internally, we call it CRS (Content Repository Service). This system supports 65M connected devices, with an average sustained load of 50K document operations per hour.
Security and compliance with regional regulations are key requirements of web services. As an example, we must ensure that customer’s data does not leave its proper legal zone. This results in the content of our global user base being spread across six application instances, deployed in three datacenters around the globe. These instances run in containers inside a set of six federated Kubernetes clusters.
In this talk we will describe the best practices learned from running globally distributed, regionally smart Apache Oak instances, including deployment across federated Kubernetes clusters, rolling upgrades, running and monitoring Oak on containers. We will also describe an open sourced data discovery and API routing mechanism developed jointly with Apigee/Google.