Kubernetes Upgrades with Low/No Downtime

There are several challenges in DIY Kubernetes upgrades. To overcome these challenges, using a managed Kubernetes service can protect your organization from environmental downtime, avoid throwing operational bodies at it, and typically automates the upgrade process.

Read the Blog: http://bit.ly/2IpveuX

Transcript of Video:

Jeremy: [00:18] I often get asked in conversations with customers, "Why should I upgrade Kubernetes?" And what I respond with, "There's a couple of main reasons, first and foremost, to get the latest and greatest in terms of feature sets you're always going to have to stay current with the latest Kubernetes releases."

Jeremy: [00:34] The second is a little more concerning for enterprises and that's security patching. As Kubernetes continues to evolve, the security aspects of the platform are also going to take leaps and bounds forward as well. And you wanna make sure you stay current to make sure that you're protecting your clusters.

Jeremy: [00:52] So, when we start to talk about upgrading Kubernetes the question turns to, how? And when you do it yourself there's a couple of key points to keep in mind, how many operational bodies are you going to have to throw at the problem? What's your tolerance for manual labor versus automation? And how do you ensure that you have data plane in high availability?

Jeremy [01:15] So, if we start to tackle those one by one, operational complexity is relatively high with Kubernetes, and you have to take an individual and manually train your nodes, upgrade that node, put it back in the cluster, and you have to repeat that over and over again until your cluster's upgraded.

Jeremy [01:33] Now, that moves us into the second phase which is automation. You can take the time to go ahead and automate those scripts but nothing exist out there today. So, you either spend your time automating, or spend your time doing it manually.

Jeremy [01:47] Then, last but not least is, how do you ensure that your services stay up and are highly available? Really, the way you do that is leveraging some of the goodness of Kubernetes, making sure you're spread across multiple nodes so you have that high availability baked in. Then, doing your nodes, upgrading your nodes one by one.

Jeremy [02:06] So, the obvious alternative to that is going with a managed service for Kubernetes. Most of these managed services provide a way to upgrade painlessly, so protect you from environmental downtime, protect you from throwing operational bodies at it, and they'll typical automate it for you.

Jeremy [02:23] For example, Platform9 allows you to upgrade your clusters with the click of a button; where we automatically drain those nodes for you one at a time, upgrade those nodes put them back in the cluster, and ensure that you have that high availability across the backend without having to throw any additional bodies at the problem.

Jeremy [02:41] So, when you start to ask, "When and how do I upgrade?" The answer should be pretty simple, identify a managed service that can take away that operational complexity for you, and really make it as easy as possible for your upgrade experience.

Jeremy [02:58] Upgrading Kubernetes is complex, when you choose to go DIY you have to solve the problem by throwing a lot of people at it and really worry about high availability. Going Cloud-native shouldn't be this complex, if you choose a managed service they typically solve the upgrade issue with the push of a button allowing you to rest easy without worrying about all that extra overhead.

Previous Video
Fission Workflows: Composing Complex Serverless Functions
Fission Workflows: Composing Complex Serverless Functions

Next Video
Maintaining VM High Availability Using OpenStack Masakari
Maintaining VM High Availability Using OpenStack Masakari