EKS Anywhere, Kubernetes protection at scale with Dell's PowerProtect Data Manager

This article is part of the EKS Anywhere series EKS Anywhere., extending the Hybrid cloud momentum

This is one article that I cherished the most as the outcomes really inspire the highest degree of confidence for Kubernetes Day-2 operations. 

Also, a big shout-out to my colleague Raghava Jainoje from Dell’s Data Protection team for partnering towards this article. Let's overview the challenges, opportunities and the strategies that Dell Technologies brings forth towards protecting Kubernetes clusters and workloads at scale

NOTE: Although we are using EKS Anywhere as a Kubernetes platform, the solution can be used in any Kubernetes distribution. The general workflow remains the same for all Kubernetes clusters and a very powerful value proposition of restoring with cross-cluster, cross-provider, cross-distribution support

For the remainder of the article, we will use PPDM as an abbreviation for Dell PowerProtect Data Manager

Context

• We have an existing EKS Anywhere cluster deployed with sockshop application and services as per the previous article
• Our goal will be to protect the sockshop application including the 8 microservices, 6 data services, TLS secret and the ingress resource definition
• In doing so, PPDM will safeguard the entire namespace of sock shop and restore the same upon any eventuality
• The validity of PPDM data protection will come from verifying sockshop application state after recovering from a complete namespace failure
• This goes beyond verifying that the pods, persistent volumes, etc. are restored
• We will insert some testing data into sockshop and execute a namespace failure.
• Upon restoration we will validate if the testing data including user logins, order and shipping history etc. is maintained

The entire act of protecting sock-shop on EKS Anywhere with PPDM has been captured as a video. This includes a detailed execution including a fresh installation of PPDM, observing current state, backing up the cluster and sockshop application, namespace failure and finally restoring and validating the sockshop application state

The video has been adequately timestamped and in case one is interested in specific narrative, then the below should help

• 00:00 - Introduction
• 00:51 - Install PPDM
• 07:55 - Initial setup of PPDM10:18 - Multicloud value proposition of PPDM
• 13:57 - Observing initial state of EKS-A cluster and sockshop
• 19:58 - Configure PPDM
• 25:04 - Discover EKS-A cluster from PPDM
• 27:25 - Protect sock-shop namespace in PPDM
• 39:21 - Delete and recover sock-shop application via PPDM

Hope the scenario and the videos were worth inspiring the level of confidence to run production and stateful persistent workloads on EKS Anywhere or generically in any Kubernetes cluster without worrying about data loss and diasters,

cheers,

Ambar Hassani & Raghava Jainoje

#iwork4dell