An internal investigation: observations from migrating multi-tiered applications to Kubernetes

Track 6 DevOps
11:2014/11/2020

Migrating large, complex, multi-tiered applications to Kubernetes can be a challenging task. In the talk, we share the experience of migrating our service-based, unified monitoring solution, NetEye 4 (https://www.wuerth-phoenix.com/en/solutions/it-system-management/unified-monitoring/), to a more modern micro-service oriented solution like Kubernetes, with the ultimate goal of providing a monitoring solution as a Service for large and multi-tenant infrastructures

Currently, NetEye 4 is built on top of the Red Hat cluster technology, which provides a set of features to implement resilient applications. With this technology, each tier of NetEye 4 is indeed high-available and distributed across multiple physical servers.

However, the burden to obtain a high-available, highly performing and scalable solution is high: it requires custom implementations to leverage the full set of functionalities offered by the infrastructure, which is not always available out-of-the-box.

Can Kubernetes be the possible solution to these problems? Which complexity is hidden under the hood of Kubernetes? Is Kubernetes complexity justified to fulfill our vision of offering unified monitoring as a Service and optimizing scalability in large environments? To answer these questions, we decide to use a product-testing method to explore potential solutions.