The main reasons for using Amazon EKS in our use were third-party solutions that were distributed as Helm charts. We were using Rancher to manage multi-cloud deployment for unification. We are also using it for evaluation purposes, building customer pilots and prototypes. Sometimes it is easy to make the build chain run through and come out as images and deploy them into Kubernetes.

It completely depends on use case. If you have got a very dynamic or a requirement to scale very fast with nodes, then Amazon EKS is a very good choice because you have got that reach and the ability to scale quickly. But if you have got a fairly static load, it becomes quite expensive quite quickly. They are expensive CPU cycles.

The main benefit of Amazon EKS is its rapid deployment. The fact that we can deploy it very quickly with infrastructure as code and then tear it down again when we are finished.

There is no real advantage to us from Amazon EKS because the advantage is the fact that we have a unified management product so we can deploy concurrently into multiple clouds and on-prem out of one pane of glass. That is the key thing there. As far as the development and presentation, sometimes it is easy just to load it up through kube control, sometimes you put it through a GUI control in front.

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.

We have been dealing with it from the beginning almost, since 2019.

Only in the past I think it had issues. The fact that regions only had a single control plane left a little bit of vulnerability in there, certainly in the early days, but I do not think that matters now. They seem to have solved that.

I had no problem. It was stable. Very stable.

It was very easy to scale.

The current stuff I am working with has been Kubernetes and building out operational software using Kubernetes. I was actually reviewing Nutanix as an option for some of the stuff I was building out.

Mainly on-prem, we are doing production work with a number of customers. We support them, we run an operational arm as well. I have been involved in platforming on Kubernetes, but we happily support any variant. We are cloud agnostic. So these distributions, we would use Amazon EKS or AKS, but not for long.

The driver in Rancher, as long as I do not have anything extremely different or complex, works completely the same whether I am driving the application onto Amazon EKS or onto a local on-prem.

We have not been using the automated patching. If we were in anger, we do not run the stuff long enough in Amazon EKS at the moment. Really, it is just up in demo and then torn down again. A lot of the stuff is being driven from other automation anyway, more infrastructure as code stuff. So that actually just gets driven completely in there.

I think that Amazon, every other provider, is adapting to the changes in the market now because the major cloud benefits are now fully saturated. Nobody else is going in for those benefits. They are starting to hit the reality of regulated technologies that are high value cannot be under a single provider. So a single cloud provider is not sufficient to support critical industry anymore. You have to have either multiple cloud or hybrid just to meet regulation in the future. So that constrains some of the flexibility. But the clouds are all working towards more on-prem extension, that sort of thing to make it more feasible.

I would rate Amazon EKS a six out of ten. I have a particular penchant for not actually overscoring anymore because of the way that people use this stuff. In other words, I consider adequate doing what it says they claim it to do. So that is a five or a six as they did what they said they would do. There is nothing wrong with that. It is what we agreed. I paid for it, they delivered it. I am satisfied.

For the usual use cases of Amazon EKS, we have been running different kinds of servers, such as web pages, and we have also used it to provide the SaaS solution for the end customer, delivering software as a service to the end customers. Basically, I deal with apps, SaaS applications, and websites.

We don't use Amazon EKS internally for us; we usually provide the service to others for their solutions.

The most valuable capability of Amazon EKS is managing the management portion of Kubernetes, which is the best thing that we get from Amazon EKS. Even though we have to upgrade it every six to seven months, it provides all the things required for us, so if a person knows how to install it and make use of it, that's all that's needed for using Amazon EKS.

Regarding self-healing nodes of EKS in minimizing administrative burdens for my customers, I appreciate that this feature allows nodes to try to heal themselves, which is a good feature. However, I haven't used this feature often, but having the service's ability to find and heal themselves or replicate is a good option for Amazon EKS to maintain high infrastructure uptime.

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.

I have been working with Amazon EKS for seven years in Kubernetes only, and I would say four years in Amazon EKS specifically.

My experience with the deployment and initial setup of Amazon EKS is usually straightforward. If you have the right Terraform code or the right knowledge, you can create it from the console or using Terraform or other tools; I don't find difficulties in starting a Kubernetes cluster. However, using the right principles or tools together with Amazon EKS may be challenging.

In terms of stability, Amazon EKS is pretty good. I don't encounter issues with stability if I use the right methods of deployment. For instance, using spot instances that are inexpensive can lead to downtime if not managed correctly.

My experience with customer service and technical support of AWS for Amazon EKS has been generally positive. Earlier, I was in developer support where the response time was maybe three to four hours after ticket submission, but I was able to resolve most problems with their support. I don't have any complaints.

I would rate technical support for Amazon EKS at an eight out of ten based on my experience with the developer support.

Positive

I have been using Amazon EKS for four years. In addition, I have used other solutions.

The integration of Amazon EKS with IAM is easy; if you have the right policy in place, you can create a role from the policy and then apply it to the application that you are using. It provides a way to use IAM to provision the software and infrastructure portions, as well as integrating application users into AWS IAM, making it very easy to implement if you know how to do it.

The influence of EKS's integration with other AWS tools on application development and management processes is significant; EKS itself is just the infrastructure. Application development requires the right tools with Amazon EKS, as it only provides a place to deploy things, and not the entire development cycle or management of workstations and servers. You must use something on top of Amazon EKS to fulfill the development cycle or CI/CD pipeline. Once the CI/CD pipeline is developed with Amazon EKS as the deployment platform, it becomes easy for developers to develop and test applications in the cluster.

I am not exactly sure about the pricing of Amazon EKS, but I think it is priced at the instance level, meaning EKS itself is not that high in price. However, whatever instances are used for Amazon EKS will determine the actual costs, particularly the traffic coming into the cluster.

Currently, I am not working with any software other than Amazon EKS, but we have plans to utilize some other applications, not just Amazon EKS, involving other services of AWS.

On a scale of one to ten, I rate Amazon EKS an eight.

I am using Amazon EKS as an integrator.

Regarding Amazon EKS integration with IAM, I do not use it.

To use Amazon EKS as a cloud provider and as a Kubernetes cluster managed by a cloud provider, it offers more benefits because you don't have to configure the cluster on your own. You can use the default configuration and just set the right networking space, set the subnet, and a few other things, but you don't have to raise up or configure your own cluster.

Self-healing nodes help to minimize administrative burdens in the organization. It helps to keep the nodes up and running. Then you can use other solutions to minimize costs or to keep the nodes running most of the time.

You can use Amazon EKS to raise up clusters and deploy applications in the cluster. The cluster is managed by Amazon, so you don't have to configure it. You can use the basic configuration of Amazon, and you don't have to interact with etcd or with the Kubernetes most inner parts. It's more simple to use.

Amazon EKS can give you more flexibility to configure on their own. In general, it's a good product. There are many different products that can fit the needs of the user or the customer in every part of AWS. In Amazon EKS, they can give a class that can be more configurable from the user or expert user rather than just using the default EKS.

Amazon EKS support for AWS tools integration has an impact on application development and management because everything is deployed on the cluster. You have to debug the various pods in Kubernetes. There isn't a direct impact, but there is an impact because everything you deploy through the pipelines goes to the cluster and there you have to integrate the cycle.

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

I have been using solutions similar to Amazon EKS from other vendors. I used GKE from Google, the cluster of Azure, and I also used KMinikube. Of course, it's not the same thing, but in general, it can be compared. For testing, K3s are just different distributions. In general, I have used other cloud providers.

It's quite easy to install Amazon EKS.

In Amazon EKS, I see it as a stable product. I don't see any particular issues in terms of nodes or performance.

I believe Amazon EKS is scalable.

I have never used the technical support from AWS.

Neutral

I see return on investment with Amazon EKS. You can save in terms of time because you can raise up a cluster or more nodes, and you can raise up the storage of the particular node in a few minutes. You don't have to take care of managing the machines directly. There is significant time-saving. You don't have to take care of the rack system because AWS has a team that works that part. You have just to pay. In terms of price saving and money saving, it depends on everything, but in general, you're going to save money.

In general, the price of Amazon EKS is expensive, but it depends a lot on the user knowledge of the tool and how we can manage the cost, which solution offers better, which solutions they can use to reduce the cost. For example, the different VM types, the Preemptible VMs or the Preemptible nodes, or you can pay one time for the use, you can reserve the machines. There are many different ways to reduce the cost.

I see big differences between Amazon EKS and GKE. The way you install the cluster is different. In GKE, when you install the cluster, it raises the nodes for you. In AWS, you can install the cluster and then you have to raise up the node using Auto Scaling Group or whatever. It's more integrated maybe. Also in terms of documentation, Google is different from Amazon.

Regarding the automated patching feature for Kubernetes clusters in Amazon EKS, I don't know any patching feature.

On a scale from one to ten, I rate Amazon EKS an eight.

There are migration projects where we need to migrate some on-premise services to Kubernetes. In that case, we have used Amazon EKS to migrate our workload. There are two production services that we need to deploy on the cloud, so we chose to use Amazon EKS to deploy them on the servers.

There are multiple node clusters that we have within Amazon EKS. We have different node groups that we have used. Some of them are high performance, some have high-speed hard disks, and some have high CPU. There are multiple node groups that we have provisioned and also enable the scaling within the pod level and within the cluster level. There are multiple features that we have used along with node management.

For streamline, we need to apply GitOps within Amazon EKS. Whenever there is a commit in the deployment-related stuff, we need to deploy the new image on the image repository, and from there, we need to update the kubectl YAML files with the new image tag. We can also choose the Helm charts based, and we can also choose Argo CD for the automated deployments.

We have implemented two things with Amazon EKS. First of all, OIDC-based connectivity between the AWS services to the Kubernetes workload. Additionally, we implemented RBAC. We have used RBAC to provision IAM users to have proper security and a constrained environment so that read-only users can only read the things.

We have used Amazon EKS Anywhere for on-premises deployments, and within Amazon EKS, it has an air-gapped environment where we can deploy the things and manage the local type of Kubernetes.

Managing the production workload in Amazon EKS is highly valuable. It has many features, for example, self-healing, automatic deployment, rollouts, and better management. It is an Amazon managed service, so we need to just focus on the cloud itself.

We have installed Prometheus and Grafana within our Amazon EKS, and we have used DataDog operators as well to have proper production workload management and monitoring management system. We can see the logs, alerts, and everything related to it. We have integrated Slack and email addresses through SES. This is how we are monitoring our production workloads and get alerts based on problems within our production.

The main benefits from using Amazon EKS include it being a well-tested product that we can use to deploy our workload. Its management system is very efficient. We can deploy things very easily and resolve our issues efficiently. It has deep AWS integration and a managed control plane. It has built-in IRSA-based security model for IAM roles. We have flexibility and portability. We can use Helm, Argo CD, Flux CD, Istio, and Linkerd. We can use multiple EC2 machines, such as spot instances and Graviton-based instances that are GPU-powered. These features help us use Amazon EKS for our production environments.

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.

I have been working with Amazon EKS for three to four years.

The setup of the Amazon EKS environment is quite fairly simple, and we can easily automate them using Terraform as well.

Once everything is set and done, there is no maintenance required, which is a very good point.

I did not see any crashes, downtime, or performance issues with Amazon EKS.

We have enabled the cluster-level enablement and auto-scaling in Amazon EKS, and that helped us significantly. Whenever we have peak time, it will automatically manage the node for us, and there are more nodes to manage our workload.

For a few of the cluster upgrades that got stuck, we connect with Amazon tech support, as we have business support with Amazon. We connect with them, and we figure out the problem. They help us to troubleshoot the issue, and we run commands related to Cordon or Taint. With this, we complete our upgrades process.

On a scale of one to 10, I would rate the technical support and customer service of AWS as eight to nine. They are very technical, and they can easily fix things within 20 to 30 minutes.

Positive

I worked with on-premise Kubernetes, and I also worked with Azure Kubernetes, AKS.

I have worked extensively with Amazon EKS, and I have automation scripts ready. Whenever a new client comes to me, I request them to see my previous portfolio, and I show them my Amazon EKS work and how quickly I can set up their Amazon EKS and run the CI/CD so that their workload can get migrated to Amazon EKS. It is quite easy for me now to advocate for Amazon EKS.

We have a big business up and running in Amazon, and we have different AWS accounts: dev, test, and prod. Based on that, there is a root account who is paying the prices for us. The workload and the profit that we are getting, we are satisfied with what it is offering.

The pricing of Amazon EKS depends upon the node group that we are choosing and what kind of auto-scaling that we have implemented. It is totally dependent on those things, including what kind of network bandwidth that we are consuming.

I find the pricing of Amazon EKS reasonable. We actually are going to pay a yearly advanced payment, which helps us save on cost.

Amazon EKS has strong integration with its services. In Azure Kubernetes, it does not have strong, deep integration with its services, for example, Active Directory, Azure Registry, Azure Monitor. Both have their managed control plane and node management. Amazon EKS has a different pricing model - it is per cluster control plane level, whereas AKS is not on the control plane fleet's history. Regarding security, Amazon EKS provides IRSA-based, IAM role-based security, whereas AKS provides directory, Azure Active Directory.

It is a very well-known product, and there are many clients in the market who are using Amazon EKS, so it is the best service offered from AWS. I rate it a nine out of ten.

At the moment, whatever is required in Amazon EKS is already there.

As integrators, we are the user of Amazon EKS. Our customer's main use cases for Amazon EKS are mostly internal services automation and development and deployment automation, which we are using for the digital applications of our customers in the AWS cloud.

The most valuable features in Amazon EKS include a suite of different services, such as code versioning, pipelining, code deployment, and code quality checks; we are using the total suite of the Amazon EKS.

From a project management standpoint, the automated node provisioning feature in Amazon EKS helps to streamline the application deployment process because the benefit I see with this product is that our deployment turnaround time reduces considerably, and all compliance standards are met. Code quality is maintained, and there are specific indicators designed in the product to identify quality issues with minimal effort. The Agile mode of project execution methodologies can also be implemented due to the code version controls and pipelining controls.

Most importantly, the deployment side of code releases and release management features are maintained without major hassles, within a short span of time, and without delays to customer experience. From both operational management and project management angles, these tools address human as well as programming side hurdles, eliminating most gaps and enabling timely project execution without surprises and ensuring quick turnaround time while meeting delivery deadlines.

Amazon EKS's deep integration with AWS services such as Identity Access Management impacts the approach to security and compliance because it integrates with IAM authentication and the multi-factor authentication process, allowing access only to those who need it.

I recommend that Amazon EKS could be improved by integrating AI intelligence with its components because EKS or the Kubernetes cluster has not yet undergone an AI wrapper. An AI component integrated into Amazon EKS, such as failure analysis and intelligent recommendations when failures happen, would be very helpful. Although automation is present, this AI feature would enhance the overall capability.

I have been working with Amazon EKS for almost four or five years now.

I have not faced any challenges with Amazon EKS; it's good and quicker. Initially, implementation can be a bit time-consuming, but once set, it becomes a 'code it and forget it' sort of environment.

I have no complaints about the stability of Amazon EKS; it is good, with no concerns at this point in time.

Amazon EKS is good regarding scalability; it suits conventional services and has room for improvement with upcoming agentic AI and GenAI resources.

My experience with Amazon support is good; we always had a good association with them, so no concerns on that. I would rate the support of Amazon EKS above seven.

Positive

Customers have seen a return on investment with Amazon EKS because they are happy and see value in the services; however, as the volume grows, the OpEx cost also increases, so any respite on OpEx cost for customers with exponentially growing volumes would be helpful.

Regarding the cost of Amazon EKS, I would say it's relative to the organization based on release management processes; for big organizations where customer experience matters, such as in the retail segment, the cost can be very high due to numerous daily changes. For companies in telecom and retail, as Amazon is a pay-per-use service, our usage levels are high, so I would desire if Amazon could provide certain discounts or credits for customers who utilize a lot of resources.

Our customers use the solution on the Amazon AWS cloud. They purchase Amazon EKS through AWS only.

For those looking into using Amazon EKS, my advice is that it is a good product, although the downside is that if the volume grows, the OpEx cost increases significantly. I would strongly recommend it for small and medium companies. For big companies with more releases, it also attracts more OpEx costs, so if there is a mechanism to control costs, it can be suitable for everybody.

On a scale of one to ten, I rate Amazon EKS an eight out of ten.

From AWS, I use many services, but mostly my work revolves around Cloud Native, specifically Amazon EKS. Kubernetes is my area of focus and expertise. Most of my expertise is around Kubernetes and Cloud Native technologies. This is why I don't call myself a full cloud offering expert, but I mostly focus on the Kubernetes usage with other OSS solutions around K8S. It's not really a niche; it's huge.

I handle both application workloads and data ingestion workloads.

Based on my experience, the best features are backed up with extensive security that AWS allows and EKS is firmly integrated into their entire AWS cloud offering. The second feature is the ability to do what we call an in-place upgrade (upgrading an existing cluster), which is a very strong capability. Additionally, the ability to integrate other add-ons such as service mesh exists, but I don't use it heavily. The ability to use all EC2 node options, including GPU options, works quite efficiently.

The freshness of the Kubernetes versions is the most interesting aspect around CSP's Kubernetes offering; it's about how close they are to the latest and greatest of Kubernetes. GCP is the leaders in that area, but Amazon is quite close behind, which is very important.

It is definitely helpful to streamline the application deployment process.

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.

I am still using all these technologies.

I would rate stability as nine out of ten.

I would rate scalability for Amazon EKS as nine out of ten.

For technical support with a business plan, I would rate it as nine out of ten.

If support is not paid, I would rate it as six out of ten.

This difference is mainly because of the response time.

With business support, I rate the solution overall as nine out of ten.

Positive

Considering the pricing of the product, I think it's affordable because it's mostly about EC2 stuff, and the control plane is not too expensive, taking into account what they do behind the scenes. Managing my own stuff in an on-prem environment helps me say that it's quite inexpensive in that aspect. The control plane is cheap, but the pricing of EC2 remains the same. I mostly don't prefer using EKS Fargate - managed containerized environment because it makes the devops team dumber and allows vendor locked in; it prevents me from managing my own infrastructure - scaling and fine tuning of resources' usage. Using margate as Autopilot in GCP and other products makes me more inclined to be locked in since it provides many features without the need to think much, but eventually, that's how the CSP will lock you in.

The integrations with IAM and Elastic Load Balancing are fundamental aspects. EC2 is the most important integration, and IAM is very strong in Amazon EKS, stronger than in other clouds. However, I need to compare it regularly as this landscape changes daily. The ELB and all the load balancing capabilities are quite strong in Amazon architecture and Amazon EKS architecture as well, so it integrates efficiently. I miss the flexibility to use other options, but I understand why they integrated it so tightly into their platform.

This isn't only an Amazon issue; it also occurs on GCP and other platforms, including Azure.

Overall rating: 9 out of 10.

We are going to be using Lambda on the AWS Stack. We are migrating all our on-premises applications to AWS.

Eggplant is going to be on the AWS platform soon. We are using Lambda functions and CloudWatch for monitoring. We are also using Elasticsearch and ELK.

After we move all the applications to the cloud, we're going to move them into Amazon EKS as containers for easy management. It would be the same as how we are using EC2 instances, but in this case, we're going to move all applications to run on containers, Amazon EKS containers.

We use it for common file sharing across applications. I'm not really sure about all aspects as I didn't use it much. We just set it up to make sure all applications can read. The main capability is allowing different applications to access the same file or resources at the same time, providing collaboration capabilities.

The whole platform is easy because once you set it up, it's easy to onboard applications. It's easy to troubleshoot, run multiple clustering, log management, and perform troubleshooting. You can also install other applications, third-party applications, integration, and all that.

It helps with having dedicated nodes for certain functions. If you want to segregate the work instead of using all the nodes, you're able to align and assign which node processes specific applications, jobs, pods, or runs certain pods.

It reduces the deployment time, setup time, configuration time, and is cost-effective. It's very highly scalable.

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.

Positive

I can recommend using it to save costs and for faster deployment, better performance, security, and easy clustering. It's easy to set up, manage, and monitor because when it goes down, it's easy to identify issues. Troubleshooting is straightforward, unlike other applications where you have to do a lot of digging to find the root cause. With this solution, you can see problems directly on your dashboards.

We are migrating our services into container services. We build websites and all of our products' backends are based on Amazon EKS.

The simplicity and management portal make it a neat solution. You don't have to fiddle around with too many open source tools, as it's just a comprehensive solution.

We use the pipeline, which is critical for us to deploy automatically. This eliminates manual intervention, which is really helpful.

We initially had some issues getting the logging out of it, because what they're providing into CloudTrail is what we get. If we wanted to go in-depth, we had to deploy third-party tools. We did try the sidecar way of getting the logs. Ideally, if the platform was able to provide those kinds of valuable logs, that would be beneficial. Adding enhanced logging capabilities would be a nice improvement.

We have been using the solution for three plus years.

Time to value is good with fast deployment and very good documentation that is really helpful.

I don't personally deal with the costing part, but I think it's a fair amount. That's the only reason we're using it continuously, as otherwise we would have moved somewhere else.

The implementation was done in-house.

On a scale of 1-10, I rate this solution a 9.

The use cases for the product involve provisioning of infrastructure and auto provisioning of infrastructure.

I have managed on-premise deployments in my use case with a Helm chart.

The biggest advantages of Amazon EKS include load balancing, auto scalability, and platform integration.

The solution includes automated node provisioning features.

The integration with AWS services involves platform services only.

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

I have been selling it for almost two years.

The scalability of Amazon EKS deserves a perfect rating of ten.

The technical support from Amazon deserves a rating of ten.

Positive

I would rate the ease of installing Amazon EKS in the middle area, giving it a five.

I have moved to pre-sales activity now.

I am selling Kubernetes Engine from Amazon.

I can rate Amazon EKS as nine because I just need to see some improvement.

I want to be a reference for Amazon.

The overall rating for Amazon EKS is 9 out of 10.

Our typical use case for Amazon EKS is that we have a number of applications and microservices that we host in EKS. We have a separate code base for the infrastructure platform, and the microservice team and the application team will be deploying their microservices on their own. We have configured it in a way that it could be easily accessible for developers as well as the platform engineers; we just platformize things. Earlier, I was using ECS, and the reason we use Amazon EKS is for better adaptation of Kubernetes, fitting our multi-tenant model.

The best features of Amazon EKS are that it is very plain by itself, but we use a number of optimizations, such as Carpenter for scaling and node auto-scaling, and Keda for application and microservices auto-scaling, as an event-based auto-scaler. Additionally, we use Portainer less, and for configuration, we utilize Cert Manager and Istio. It's not only Amazon EKS but a combination of various components within it.

By default, if you just install Amazon EKS, you can deploy your application, but to have it enterprise-ready, you have to configure a number of other things that will boost productivity.

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 have around six years of experience with Amazon EKS.

Amazon EKS is a stable solution, as it is only available in AWS alone.

It is a scalable solution for us.

Before using Amazon EKS, I was using ECS. I switched from ECS to Amazon EKS because our product design changed. With numerous small services that you don't want to manage the backend infrastructure for, you can easily deploy and let it be with ECS; it is a more straightforward solution. However, considering cost with Amazon EKS, it may be pretty high, but it serves its purpose very effectively without management overhead.

If you are going with Amazon EKS, you must change your deployment strategy and develop applications for Kubernetes, writing deployments and pods, or stateful sets, which provides more flexibility. There are pros and cons to both solutions, and you have to evaluate which will suit your use case. In our situation, we had some applications in ECS as in Amazon EKS, and that was an architectural decision discussed internally within teams.

The initial setup with Amazon EKS was hard initially, but being accustomed to it now, it's not that difficult; it's relatively easy.

We have seen ROI with Amazon EKS; we have a separate team actively working on it. We have cost explorer available, and a bill forecast based on usage allows us to determine whether resources are underutilized or overutilized. You can generate reports and analyze them. I have done this for ECS, but for Amazon EKS, I haven't worked on cost savings directly, as there is a separate team responsible for that.

My experience with pricing for Amazon EKS is limited as there's a separate team for that, and I do not have much knowledge of specifics. However, the pricing is based on the instance type we use in the EKS node group, so it should cover that aspect; their pricing is generally easy to understand in terms of instances.

We are using a cloud deployment model. On a scale of one to ten, I rate Amazon EKS an eight.