Amazon EC2 Auto Scaling vs. Amazon Elastic Kubernetes Service (EKS)

If you need to establish a system right away, in the past it took weeks or months to request a quote from the vendor and receive the equipment. Now, with Amazon EC2 in less than tens of minutes or hours, you can create a test environment and test it without any inconvenience.

It is well suited when you want to have a Kubernetes cluster in AWS Cloud and want to avoid all the management overhead of maintaining your own cluster in terms of the control plane. EKS seems to be lacking in features when compared with AKS and GKE. Backups, service mesh, and monitoring have a lot of room for improvements.

Incentivized

• Dynamic scaling can be configured to respond to a wide variety of metrics and alerts
• Predictive scaling allows one to get ahead of high traffic events rather than simply reacting to them
• Health checks are configurable based on the needs of your application and architecture

Incentivized

• Upgrade the kubernetes clusters to the latest version with a single click
• Auto scaling policies to automatically scale the nodes
• Detailed logs and events on the cluster within the EKS clusters portal, cloudwatch logs and metrics

Incentivized

• Enable cross region auto-scaling. It currently limited to a single region.
• Support custom health check scripts, deeper integration with application-level metrics. At present, health checks are basic.
• Adding cost-aware scaling scripts that factor in instance pricing and recommend optimal configurations based on budget constraints.

Incentivized

• AWSIAM integration with Kubernetes RBAC could be better.
• Enabling some add-ons like service mesh, and monitoring will be nice instead of having to install them yourself after the creation of the cluster.
• EKS bootstrap time could be faster ...

Incentivized

Usability is good since we already know how AWS works. For those that are new it might be a little bit confusing at the beginning but they are improving it at a fast pace. Even though AWS keeps changing the user interface constantly, it is still powerful, understandable and easy to use. For technical people, they still offer the CLI.

Incentivized

Cluster maintanence is reduced, easier to deploy resources, great observability insights

Incentivized

The platform works as is. The help and tutorials on the help page can help you to setup the entire platform without problems, and also provides help on a huge variety of problems. Amazon also provides support plans. We have the basic support plan, but Amazon offers three support tiers, and we know that it works perfect.

Incentivized

The main reason is our total infra is created on AWS and we tend to use the natural service by AWS rather than third party tools, which has more advantages when the auto scaling interacts with other AWS services and its way easy to configure when we compare it with counter parts like Autoscale from Microsoft Azure.

Incentivized

It feels like AWS is behind the EKS race, the only advantage I'm able to see right now is the support of IPv6, however, trying to promote AWS alternatives that are different from the market and more like a vendor locking solutions like ECS/Fargate have kept AWS behind and focusing on the wrong things. EKS needs to really improve its integration with the Kubernetes ecosystem and have an enterprise solution for monitoring, backups, and service mesh.

Incentivized

• We will devote more time to development than server administration, but we will require additional time if you migrate from another ecosystem.
• Fault detection and reporting are automated in the old server, and bandwidth is fixed per month, but everything is manageable automatically. We only pay for the resources we use.
• After some months, we met our return on investment (ROI).

• Good performance of platform without hiccups
• Less number of people required to manage cluster
• Easier to deploy new microservices

Incentivized