Container-Native Storage Performance

To read this article from Brookend Limited sponsored by Ondat, click on:
Container-Native Storage Performance in the AWS Public Cloud – First Edition

Container-native storage (CNS) provides an elegant way to introduce data persistence to containerised applications. The technology has been available within Kubernetes clusters for approximately five years, with a mix of commercial and open-source solutions in the market. Data persistence is essential when running stateful applications in the public cloud, where the resilience of individual virtual instances is not 100% guaranteed.
The public cloud, and specifically AWS, offers several solutions to map persistent storage to virtual instances. The characteristics of these storage offerings vary based on resiliency, capacity, performance, and cost. As customers deploy container-native storage solutions into the public cloud, choices must be made on which storage solution to use.

Firstly, bear in mind that container-native storage solutions offer more than just resiliency. CNS deployments provide abstraction, data protection (snapshots, replicas), data optimisation (compression, de-dupe) at the container level. Using storage classes, an application could be deployed across any public or private cloud, without needing to understand the underlying storage solution.

However, the specific storage solutions in place do matter. AWS and other cloud storage providers have designed their storage offerings with specific use-cases in mind. These features affect the performance and crucially, the deterministic nature of containerised applications deployed on top of cloud storage.

When deploying container-native storage, system administrators and IT users could simply choose the default option and use block storage connected to virtual instances. As we will show in this report, the choice of storage for CNS is affected by storage performance, cost, and resiliency. We will dig deep into the details of the implementation of each storage solution, helping the end user pick the right choice for their application. This report isn’t intended to show that one choice is better than another, but rather allow the reader to make an informed choice on how to configure a container-native storage environment on Kubernetes.