What needs improvement with Amazon EKS?

Amazon EKS can be improved through enhanced security and optimized costs with better streamlined operations. Performance through automation can be improved, and the integration with other AWS services can be enhanced. Regularly adjusting the size of virtual machines and setting precise Kubernetes resource requests and limits can help avoid over-provisioning. Implementing autoscaling tools such as Carpenter or Cluster Autoscaler can automatically scale nodes and optimize resource allocation for cost savings. A separate cluster can be used for different environments to prevent accidental disruption. Monitoring and log capturing can also be improved.

In my experience with Amazon EKS, I have not encountered aspects that require improvement, as AWS has invested intelligently in its development, especially with the addition of Karpenter. I do not have any particular improvement feedback for Amazon EKS or Kubernetes at this time.

I have experience with Azure, and in comparison to Azure, a downside of Amazon EKS is that even if you want to deploy a dev workload or do some experimentation, we have to pay the charges for the control panel with no free option. Additionally, I have faced many issues while configuring the node groups and the whole configurations; bringing up the nodes was a bit hectic, and I was not able to determine which node was failing and for what reason. Specifically, the pricing for the control panel of Amazon EKS is hefty, and there is no cost-cutting that can be done on that side.

Improvements could include better support and pricing, which is always important. There are definitely areas for improvement with Amazon EKS.

When we need to deploy the application, we require a large number of instances. Therefore, I hope and believe I will not face out-of-capacity issues in AWS, especially since I have not yet experienced traffic around 50,000 plus, and I believe I will not face such issues in Amazon EKS the next time we deploy with a large number of nodes and worker nodes.Additionally, the upgradation process of Kubernetes rapidly rolls out new releases, so it should be easier for our production environment to upgrade Amazon EKS clusters. Sometimes when we are going to upgrade the Amazon EKS cluster, we need to check the backups, and we should have options to export our configurations, such as exporting the configuration to S3 or somewhere else to find backups. Other tools, such as Velero, provide this functionality to back up configurations, so I hope this backup process will help us fulfill our backup policies and other requirements. For the pricing aspect of Amazon EKS, one specific issue arises when we deploy applications, especially as we provide SaaS services to our clients. We would like to know the cost for each customer, but we face issues because AWS charges $70 USD for the Amazon EKS engine. We struggle to divide the worker nodes' fees and the engine cost among clients, as some users have low traffic and visibility while others have large amounts of visibility and traffic. Thus, we face cost-related issues when running multiple customers on the same Amazon EKS cluster.

The costing perspective could be improved. With normal on-premises servers using Kubernetes services, the costing might be slightly higher compared to other Kubernetes vendors. If the pricing becomes more comparable and they match other vendors while providing more flexibility, it would be more advantageous for AWS.

The initial setup in Amazon EKS is complex, especially compared to services such as ECS and Fargate, which I worked with in my US Bank project, involving VPC networking, IAM roles, or node groups. However, once set up, the deployment becomes easy. Using infrastructure as code, such as pipelines, I usually automate cluster creation with Terraform and integrate GitHub Actions. We use standardized Kubernetes manifests that make spinning up and scaling clusters much easier, so while the initial setup is complex, networking and IAM integrations make deployment and scaling smooth and easy to handle.

In my opinion, Amazon EKS could improve its plugin management; while Amazon offers some improvements, they often come at an extra cost. Compared to Google, for example, having the same cluster in Google Cloud is much easier because it hides all the complexity of managing Kubernetes add-ons and plugins. EKS is improving in this area, but it could still be smoother and better. Upgrade management is also a constant concern because EKS frequently forces users to upgrade, which can be inconvenient due to the risks involved in introducing something new, especially at the platform level.

From my perspective, Amazon EKS is quite sufficient. The main issue is that Amazon EKS only has an update for Amazon Linux 2023. The challenge occurs when switching to a new operating system. Currently, I am using Amazon Linux 2, but according to AWS information, it will be deprecated. I hope AWS continues to support the operating system for Amazon Linux 2.

From a DevOps engineer's perspective, the IAM roles system could be improved. When creating a cluster, we need to create and assign roles. If this process could be simplified, especially for users with limited AWS knowledge, it would be beneficial. Currently, users need to understand IAM fundamentals, how it works, and how roles and users are defined. It would be helpful if Amazon EKS offered ready-made templates, similar to AWS's default VPC option. This would make the service more accessible to new users while maintaining its functionality for experienced users. We have not encountered any issues with Amazon EKS's automated patching feature for our Kubernetes clusters.

One area of Amazon EKS that could be improved in the future is related to the behavior of the cluster autoscaler. We need enhancements so that scaling in and scaling out happens more effectively. Currently, the cluster autoscaler lacks visibility on the Kubernetes labels of a node group. The cluster autoscaler should at least be able to check the labels of the Kubernetes nodes, which is a necessary feature.

The main area for improvement in Amazon EKS relates to master node control. When setting up a Kubernetes cluster independently, you have access to the master, but with AWS, you do not have control over it.

I have one suggestion for Amazon EKS. When working on microservice architecture, we need to use that context and have K9s installed for a graphical user interface to check pods and nodes in a clear-cut manner. If that support is added within Amazon EKS itself to check all the config maps and everything within the UI itself, that would be very beneficial.

We haven't fully utilized self-healing nodes in Amazon EKS yet. The self-healing nodes are still in our research and development stage. It is new we are using in production clusters.

One area I observed during setup was that while managing it through CLI and Terraform, there are many possibilities for setup and infrastructure updates. However, I believe the console experience could improve. In the AWS console, when trying to set up an Amazon EKS cluster, there were limitations on certain features I encountered a few months back while checking. EKS frequently updates, so I don't know if there's a new release. However, I found some features that I could not manage through the console, requiring me to use CLI or Terraform. It would be beneficial if we could have all features supported through the console, providing full management capabilities there.

We have not been using it from the point of the application using the IAM. We have been using it because quite often our customers are tied back to usually Entra ID and things like that. The only concerns I had with Amazon EKS were related to cost, the usual problem you have with cloud. It is fine if you can exploit it for dynamic loads, bring it up, get rid of it again. That is where its strength is. You pay for that premium, but as far as running the thing under constant load, it is a very expensive way of deploying. In the early days, there were a couple of vulnerabilities exploited from the single control plane per region. So there is nothing stopping me deploying multi-region, and that means multiple control planes. So I could deal with that, the infrastructure handled the criticality. The only thing that I could possibly run into a problem with, which I have not had to at this point, but architecturally, is with regulated technologies, banks, that sort of stuff where you cannot be single provider sensitive.

The area of Amazon EKS that could be improved is the development cycles because every six months a new version of Kubernetes is launched. Working in the infrastructure, I have found it quite difficult to keep up with infrastructure updates and new versions. Migrating the whole infrastructure from one Amazon EKS cluster to another is quite cumbersome, so if possible, version management should be made much easier than it is now, perhaps with some option to deploy code using blue-green updates. Creating a clone of the current infrastructure, updating it to the latest version, and terminating the old infrastructure would be a great enhancement. I haven't used Amazon EKS's automated patching feature, but I believe it involves directly upgrading the AMI versions remotely or from the Amazon EKS dashboard. I understand it is a good feature, but I haven't used it yet.

The main problem or area for improvement is flexibility in configuration. This is the only concerning part and nothing apart from this.

I would appreciate seeing integration between the other services in Amazon EKS. For example, there are now more managed Prometheus and managed Grafana services available last year. I would appreciate knowing more about how AI-based services get integrated with Amazon EKS and what kind of AI automation, healing, or troubleshooting services Amazon will introduce in upcoming releases.

Regarding the flexibility part, if I want to use something such as Kong/Linkerd service mesh or other solutions, most of the CSPs bind you to their own solutions rather than allow other options to be made and integrated with, and this is what I mostly miss in their part. In terms of built-in observability in Amazon EKS, I know it's mostly about the great integration with AWS itself. When I want to integrate it with any Grafana or Prometheus solution within AWS, things work efficiently with IAM. However, when I want to cross the boundary of the CSP, that becomes an issue. Integrating some open-source solutions works, but I need to work really hard for that. The major area for improvement I've seen involves deep diving into one CSP with no equivalent solution elsewhere. The most important consideration is about the pricing and the flexibility of moving and building a multi-cloud environment for most customers. The issue is that when CSPs try to lock you in, flexibility becomes the most critical aspect. Amazon aims to put you in a very Amazon-centric environment, but you need to be aware when you're using Amazon EKS that you're not locked in. The major paradigm for customers with maturity in using cloud solutions is to avoid vendor lock-in. Most early adopters understand this, but the main mass, such as the banking companies I work with, aren't in the same state of mind; they need to build everything from a Cloud Native perspective with as much OSS as possible.

The main challenge is at the beginning when you're still learning how to set up. The initial setup and knowing how to do it the first time can be challenging. We had some challenges with the network pods. There were several pods that we had to try plugins for network, and the load balancing for the application we were deploying was quite difficult to manage. The documentation should be easier to understand for beginners. It needs to be straightforward and easily grouped together because information tends to be scattered. Some features required a lot of command line changes, especially setting up ingress and load balancers. If there could be a way to do that on a GUI, that would be easier. Having more configuration options on the UI, especially for setting up load balancers and ingress controllers, would be beneficial.

We usually get deployed and only need to tweak the source code; however, I think the monitoring part and observability part could be improved.

Amazon EKS's deep integration with AWS services, such as IAM and elastic load balancing, has created some challenges. For example, we have something in place already, and there are some issues with enabling FIPS, which is FedRAMP compliant for the load balancers. You cannot change the SSL policy for the load balancer; I am not sure if it has been patched by AWS yet. However, apart from that, we use it effectively, and it is more flexible. Regarding built-in observability in Amazon EKS, there is CloudWatch and CloudTrail. However, you cannot profile the applications; we can collect logs in S3, but there is no streaming solution available. Only CloudWatch exists, so we use other tools for observability and do not depend solely on CloudWatch, only relying on it for crucial workloads and infrastructure logs. Amazon EKS can be improved by having the maintenance of Kubernetes versions managed better, as everything is handled by the Kubernetes team and possibly a separate team at AWS. We have to constantly maintain upgrades and ensure EKS add-ons are up-to-date, requiring us to upgrade the Kubernetes version and releases. They could provide a managed service in the backend instead of making customers handle it; we are currently doing it, but it's a regular activity we do per quarter.

I am not the right person to ask what could be improved in Amazon EKS to make this tool better for the next release. A continuation of the managed pieces would be beneficial because there is no integration of clusters. They are all separate with no real managed cluster type of capability.

I would like to see improvements in the management capabilities that are native to AWS ( /products/amazon-aws-reviews ) for Amazon EKS. Our current use involves using a different tool to administer the management, and I believe enhancing this aspect would be beneficial.

There is room for improvement in making Amazon EKS ( /products/amazon-eks-reviews ) less error-prone when writing on the YAML file. A UI could help generate config files, simplifying the process for developers who are not architects. It is very time-consuming, and companies often expect developers to do everything, which can be overwhelming.

* EKS incurs an additional management fee ($0.10 per hour per cluster) along with EC2 or Fargate costs.May be expensive for smaller workloads compared to alternatives like AWS ECS.Requires expertise to configure and manage Kubernetes resources effectively.Networking (e.g., setting up VPCs, subnets, and service endpoints) can be complex.Simplifies managing multiple Kubernetes clusters, especially for organizations with hybrid or multi-region setups.Integrated dashboards for Kubernetes metrics, logs, and traces.Simplifies observability without needing third-party tools.

Sometimes I have trouble because, in our corporate network, there are various networks, etcetera. It's difficult to connect to some of the clusters, and it's easier to go through the UI when troubleshooting something. At some points, the UI seems limited to me with the functions it provides. You can get information like what kind of port is running on the cluster, but I haven't really explored the UI so far, so it's difficult for me to see the logs, for example. Or sometimes, you are only limited to the basic Kubernetes things. We have certain customizations installed in the cluster, and for that, you really have to use kubectl from the command line. You are not able to use the EKS UI to list certain custom resources. So maybe there can be some kind of improvement, but maybe it's just me that I haven't really explored the UI that much.

I would like to see a warm-up time for AWS Fargate, similar to what GCP Cloud Run has. This would improve internal security. I would also really love to see lower costs compared to other cloud vendors. AWS can get quite expensive.

The solution's graphical interface is not the best. It could be better in terms of enabling some integrations or managing the configuration.

The main thing to improve with Amazon EKS is the price. However, these services can be very expensive. For example, in countries like Turkey, the cost is too high. That's why we offer our cloud solutions locally. We developed hybrid solutions, but their prices are still very high.

They could add logging features. At present, we use external tools to increase and decrease the number of instances.

When it comes to Amazon EKS, there are IAM permissions and RBAC. When you create an IAM user, you give the privileges on the cluster level, but there won't be anything inside the clusters. In the clusters and their respective files, you will have to map the IAM user created with the help of AWS. The documentation part of the product is an area of concern that needs to be made easier from an improvement perspective.

Assigning roles and responsibilities to interact with a created cluster as a user over a command prompt is cumbersome on AWS. Initially, we create a user to interact with a cluster. Since everyone can't use the cluster, we need to assign some permissions to that specific user. It is very cumbersome to assign permissions to users to interact with a cluster. We always get errors, and it takes many days to resolve that permission issue before the user can start interacting with the cluster.

I'd like improved traffic handling and additional application details within the system.

The product’s pricing needs improvement.

If you compare Amazon EKS with OpenShift, the latter provides users with a solution that is fully managed through automation. Amazon EKS is not a solution that can be fully managed through automation, making it an area of concern where improvement is required. Amazon EKS should be manageable through a web portal or web interface, a feature that exists in OpenShift. Amazon EKS should be available as a fully managed service since we use Helm chart to deploy the product in our company right now. The initial setup phase of the product is an area where certain improvements can be made.

There is room for improvement in stability. I faced some problems with the App. The problem is actually the app, with the different teams fixing it.

I would like to see a cloud setup bank management feature.

The management of the nodes in Amazon EKS should be improved. AWS should reduce the constant upgrades of the nodes of Amazon EKS or EC2 machines.

I'm having difficulty getting my AWS clusters to communicate with my local machine.

I am not impressed with the tool's Amazon console. It also needs to add security features.

It is challenging to host the applications in the existing VPC of the solution. They should include some essential configuration features to it.

The tool's setup is complex.

Amazon EKS provides very minimum information during the upgrade of the node group. When the upgrade doesn't work well, it doesn't give enough information for us to troubleshoot. So it would be great if Amazon EKS provided more information in such cases. Amazon EKS should enable some AIOps.

An area for improvement in Amazon EKS is the user experience. The platform could be more user-friendly. Only an expert can manage and use it.

The graphical user interface could be better.

They should improve the security and include some cloud-native project integrations. In addition, they should enhance the Jira integration.

We have problems with setting up virtual environments and installing the right packages. I believe the initial setup could be a better experience and faster customer support.

The solution could be improved by adding monitoring, filtering, and logging capabilities to its current CloudWatch features.

The solution should include a popup for clusters so that all relevant information is visible at the bottom of a page. When clusters exist or are running, there isn't much detail on the first phase so navigating and clicking on different options is required to search for relevant information.

The main area of improvement is that a cluster is required on-premises, which takes a lot of time. For example, we must drain the total nodes during an upgrade from version 1.21 to version 1.22 with on-premises. After draining the total nodes, our container will shut down, and it will be recreated after upgrading. But with the AWS Kubernetes Services, the upgrade from version 1.21 to 1.22 is completed with one click. It's straightforward for the users. In any secure services, nodes are working on the EC2 services. Whatever the EC2 services, the specified AMI is available. This AMI is an auto-security package that is automatically upgraded per the company's need. It is also secure.

The connectivity could be better.

There isn't something that is unique or outstanding. I'd like to see the solution add a service catalog.

Amazon EKS is predominately public. However, the government has started to have a lot of interest in Kubernetes and is receiving more education on Kubernetes and Amazon EKS. If we can have the security of Amazon EKS align with the security that is set out by the government it would be much better.

When we switched to EKS, historically it wasn't good. There were issues with bugs in it. They didn't have managed pools, which means small subsections of the clusters that you divided into pools like a mini-cluster in your cluster. However, now they have managed pools. For the last several versions, the issue was with their kind of networking plug-in, the security plug-ins, and things like that. That EKS layer on top of the Kubernetes, they add themselves to each cloud, however, only with fewer standards and a little more issues. They need to work on the Amazon plugins on the Kubernetes cluster. We just updated to a cluster 1.18, but we were on that cluster 1.13 which had many bugs and issues. Moving up to 1.19 in the middle of last year, we had some issues which they had to fix. One thing that is probably not the greatest in Amazon is the ideology. They really want you to stick to cloud tools. They want you to use the managed version of the databases and our preference is to use the Kubernetes-managed databases. This doesn't fit well with the AWS philosophy, which is then passed on to the AWS engineers and they push that, push ideology on us as well, saying "You know what, we want you to use this database." We're not dogmatic. If they want us to use a specific database, we use it, as the cluster is very dynamic. We don't need to deploy a database within a cluster, we can use the cloud database. To us, it's just a connection string, so it's not inefficient for us. It's just based on the client. However, you can see there's a little bit of an ideology dogma baked into the AWS philosophy just to keep you in the cloud.