Azure Kubernetes Service (AKS) Reviews

We have two primary use cases of this solution. One, we deploy microservices for our customers using the Kubernetes cluster. For example, we have deployed Kafka on Kubernetes as well as Azure functions, which uses the same Kubernetes cluster. 

The other use case is for our internal usage. We only have microservices and Azure Functions, we don't have elastic or data. We do have a tight integration between Azure services and Kubernetes. The concept is called KEDA, or Kubernetes event-driven autoscaling. This means that if there are no services messaging you, Kubernetes usage will be zero. However, if there is a burst of traffic for my services, Kubernetes will create accordingly. 

The most valuable feature of AKS is that when I am sitting at the cluster, I can ensure compliance. Compliance is easy right out-of-the-box with integration to Azure Security, Azure Active Directory, and Azure Policies. The entire cluster is secure by design and not added as an afterthought. 

The other feature I like about AKS is that because of ransomware attacks, Microsoft has released specific guidance on threat matters. AKS is based on framework. This framework uses the proper guidance architect to tell you how the cluster should be designed. 

Another valuable feature is that AKS tells you exactly what you should do with each threat. It will guide you to improve the cluster, deploy something on the cluster, operate the cluster, and when you are trying to heal the cluster. It is very actionable.

Unfortunately, when a microservice fails, Azure can take up to 60 seconds to broadcast an alert to the monitoring agents. For a lot of the microservices, this is too late, particularly for serious failures.

Microsoft would benefit by offering enhanced learning and certification. Kubernetes is now mainstream and becoming more and more integral. Creating an Azure Kubernetes certification would ensure there are specialists in the market that could adapt best practices in a more structured fashion. 

I have been using Azure Kubernetes Service for three years.

The solution is stable. In the past three years, I have never experienced any serious problems.

We have internally two people to maintain AKS and for our customers, we have multiple people. There are a couple of DevOps resources who are trained in Terraform and are Azure DevOps certified who operate the cluster. 

We rely upon monitoring and alert mechanisms. Basically, we do not need to be on the server with a dedicated resource. We only need to look at the alerts, as the solution is self-healing. 

Scalability depends on how the architect designs the cluster. Once a Kubernetes cluster is up and running, the auto-scaling works according to the basic design. If I need a different node pool, then I need to create a design cluster. If this is done correctly, scalability by design or architecture is not an issue. It is up to the individual architects to be disciplined in considering future needs and build in flexibility.

The initial setup of AKS is a breeze. Straight out of the box, with experience, it takes only 30 minutes. Setting up a secure cluster takes only 30 minutes. I use Terraform extensively, so it is pretty easy to set up. I would rate the initial setup a five out of five for ease of setup.

I did the deployment of AKS myself with two other people. One from networking with an infrastructure point of view to handle the set-up of the V-Net by controlling the IP address range. The other person is the one who manages Azure Active Directory.

There is an initial infrastructure cost, however, the more services you deploy, the cost per service comes down drastically because you are leveraging the underlying infrastructure. As you scale your operations, AKS becomes more cost-effective.

The cost of AKS is humongous, it is very expensive. You are charged based on the nodes, the networking, and the storage. Multi-clusters are another consideration, especially if you need to add firewalls.

It is important to keep in mind that the cost is not just the licensing, but also the human resource cost. Every organization should have some idea about what they should be putting into Kubernetes together with the type of people, and types of resources required to monitor, alert, and operate the solution.

Compared to Amazon Elastic Kubernetes Service and OpenShift, Microsoft Azure has very specific guidance on how to secure an entire cluster with multiple inputs. This is really phenomenal. This guidance is missing in the other products.

Anyone considering AKS as a solution should be sure they consider the design carefully and give a lot of focus to the initial setup. 

I also recommend that the organization is clear on what its end goal is, otherwise they will incur too many additional costs. 

Lastly, keep in mind that it is difficult for the executive level to understand what Kubernetes is and that it is not the answer to everything.

Overall, I would rate Azure Kubernetes Service a seven out of ten.

It is a hosting platform for creating web applications and deploying them. 

The serverless capability and auto scale feature are the most valuable. 

More control over Infra scanning can be introduced. 

I have been using Azure Kubernetes Service (AKS) for ten months. 

The stability of the solution needs improvement. I rate the stability a six out of ten.

It is a scalable solution but can be further improved based on the user requirements. Presently, a few hundred users are using the solution. I rate the scalability a seven out of ten.

The initial setup is easy. Some log maintenance is needed. I rate the setup an eight out of ten. 

The pricing of the solution is same as Azure Stack. 

It really depends upon the usage. If they really have the use case to go cloud, they can use it, but still, it is more applicable for people who want to get into software engineering and web development. Especially in the financial domain or e-commerce, where you would be relying on websites. 

I rate the overall solution a seven out of ten. 

I am using AKS on a Cosmos DB database. AKS is similar to AWS in stability and scalability, but when used with C#, AKS is easier to deploy and better to use because I use .NET applications.

The documentation makes it easier to deploy AKS, so it is the most valuable feature.

The onboarding process could be improved. For example, when you open Azure's main port, it is not very easy to access the servers because you need to put your mail and password, and sometimes you need to configure the authenticator. Depending on the case, this could be a problem.

I have been using AKS for two years, and I am using the latest version.

The solution is stable.

The solution is scalable.

The customer support has been good.

The initial setup was very easy.

I have not seen a return on investment since I have only used the solution in my company for two months.

The solution is not very expensive. It is expensive when used on the Cosmos DB database.

The solution has good documentation and is easy to use and deploy. AKS is a good place to start your business. I rate the solution a ten out of ten.

Our use case is for any microservice that requires a Kubernetes platform for deployment. But primarily, my reason for using it was to develop a tool for our clients to migrate their deployments from Pivotal Cloud Foundry, which is now out of support. I wrote a tool to automatically convert Pivotal configurations into Kubernetes and then deployed them on Azure Kubernetes Service for testing the mapping and tool output.

So it's more of a migration from Pivotal Cloud Foundry to Kubernetes and deployments, with Azure AKS as the target platform.

For most of our deployments, we utilize Azure DevOps pipelines. It provides a seamless experience with a fantastic console and user interface. You can configure the pipeline steps, including pulling code from the Git repository, triggering execution, building the container image, and pushing it to the Kubernetes clusters. All of this can be done on a single screen, making configuration easy and intuitive for developers.

One area that could be improved is the Azure CLI. It would be beneficial if they could abstract some of the complexities related to deployment scripts and make them a part of Azure CLI. This would help reduce the scripting required in the DevOps pipeline for deployments.

In future releases, it would be great to have more container management functions available directly on Kubernetes. For example, auto-scaling features and the ability to configure the initial number of container replicas and memory allocation from the console itself. This would minimize the need for developers to manually modify the manifest file most of the time. It could provide an additional layer of protection and security, allowing only authorized personnel to make changes to the manifest, while others can modify configurations through the web console for specific services when necessary.

I have been using Azure Kubernetes Service for a year.

There are no such concerns or issues with stability. Any issues arising from the written code or application design may require troubleshooting.

From a platform perspective, there are no concerns on the infrastructure or platform side where we need to contact anyone.

Azure Kubernetes Service is pretty robust in terms of scalability and auto-scaling fixes.

Even large organizations are putting their workloads on AKS. They have data centers globally distributed, which is a desirable feature for many of our clients operating in multiple geographies with their businesses. The service provider has data centers distributed across Shield, offering extensive coverage. For example, earlier in the UAE and the Middle East, AWS used to be the only option, and they have strict regulations around data privacy and protection that disallow data to move outside the borders. Microsoft has set up data centers in Bahrain and Dubai exclusively for Azure, offering an attractive alternative. Most of our clients from NetRegen, who had the AWS option only, are now moving to Azure for its enhanced capabilities.

I rate the solution’s scalability a nine out of ten.

Azure Kubernetes Service is better than Amazon and Google because it is extremely seamless, and its user interface is very intuitive and minimal. There are no additional configurations that you need to do. For example, Amazon and Google are very cryptic with many indirections, making their documentation hard to read. In contrast, Microsoft has done a fantastic job simplifying everything and keeping only the bare minimum. The documentation is also quite good to support the developers and offer a better user-friendly experience for them.

The initial setup is pretty simple.

You would probably need developers and DevOps engineers for the initial setup. It is intuitive and has good documentation. So, you don't need to contact the vendors or any external consultant.

Running workloads on Amazon or Google are more expensive than Azure Kubernetes Service.

I rate the solution’s pricing a nine out of ten, where one is high and ten is low.

Azure Kubernetes Service is my go-to and preferred platform over Amazon and Google.

Overall, I rate the solution a nine out of ten.

I use the solution for infrastructure services on platforms. It can be used for running virtual instances on Azure.

Integration and automation are the best features of the solution. It is interesting to see what it could deliver by integrating all the services.

The product should make the initial setup easier. The product should provide an online format where I can see, read, study and research it on demand.

I have been using the solution for around five years.

I don't see any stability issues with the multiple servers integrated into Azure.

Scalability is one of the top benefits when you migrate to the cloud. We can go directly from 20 users to 10,000 users.

The initial setup is complex.

Nothing in Azure is as easy as plug-and-play. Most of the time, the product is cloud-based. There are on-premise solutions for very specific cases. As a representative of cybersecurity operations, I have to do research and read the manuals. Overall, I rate the product an eight out of ten.

The most valuable features of AKS are scalability, deployability, and automation.

Configuration management and troubleshooting performance issues are difficult to solve and could be made easier. 

I have been using Azure Kubernetes Service (AKS) for approximately one and a half years.

There are some performance issues with the stability. There is some fine-tuning needed.

I rate the stability of AKS a seven out of ten.

The solution is scalable.

Small to medium-sized businesses are best suited for AKS.

I rate the scalability of AKS a nine out of ten.

The support is excellent.

I rate the support AKS an eight out of ten.

Positive

The initial setup can be complex and took a total of four weeks. However, the incremental deployment took a couple of days.

In the first step of implementation, we do an estimation and identify all of our independent modules and packaging needed.

I rate the initial setup of AKS a three out of ten.

The price of AKS is expensive. We pay approximately $10,000 monthly.

My advice for others is to focus more on packaging the individual models and cables before going to where the solution is situated.

I rate AKS a six out of ten.

The requests get processed within seconds. Earlier, it used to take around four to five minutes. 

The product has  built-in functionality that checks whether the service is available or not. In case the service is down, the tool will create a new instance by default. Hence, the web API will be always up irrespective of the server or the situation. 

The product needs to support a UI dashboard. I have to execute every single command to check the status of services which takes time. 

I am using the tool for three years. 

The product is not stable. 

The tool's scalability is high. You just need to mention the maximum and minimum replicas that you need at a certain time. It will increase the load by default and upscale immediately. My company has around 200 users for the product. 

The product's setup is difficult. If everything is written in your code or you are replacing the values or replacing the credential on the run time, then it's not difficult. But if you want to read those values from the keyword, it can be challenging. The product's deployment takes only seconds to complete. We have seven resources for the tool's deployment. However, the configuration part takes time which can be around five to ten minutes since we need to create a new AKS resource, public IP, etc. 

The product's deployment was done in-house. 

The product follows a pay-as-you-go pricing model which is good for small enterprises. You need to pay only for the services that you use. 

I would rate the product a nine out of ten. The solution has good integration with ACR and .NET. Its documentation also gets updated frequently. 

We used the solution to integrate banking pipelines for our customers. Also, we integrated Kibana into AKS, and from there, we deployed their parts. The solution's features, like machine learning models and analytical tools, have also helped us.

The solution's valuable features include in-built volume controllers, integration, container orchestration, monitoring, PSI drivers, Github, Gitlab, CI/CD, etc.

There are a lot of features that should be included with AKS. Also, the solution is still growing; thus, many things must be improved. We are still testing the solution, but they should provide better monitoring services. Along with it, they should include in-built DR features.

We have been using the solution for two years.

It is a stable solution.

It is a scalable solution.

We contacted the solution's customer service during the time of deployment. We also never needed the tech team's support, as the deployment is straightforward. The event architecture was required on our end; Microsoft doesn't help there. They provide basic help to establish the SaaS components.

I used Azure Virtual Desktop and Azure Migrate as well. Mostly, customers move their applications to AKS from the premise as they want active manageability of services. The other reason is credit plans. Customers prefer switching to services that offer them credit cards. Also, the pricing matters a lot. Even though the services are performing well, people opt for vendors offering a particular cloud that is low on cost.

The initial setup for testing purposes was easy. We just had to set some commands with a click, and it took around ten minutes to complete. During the final deployment, we required a code. It is a day long process.

The solution's pricing and licensing are structured on the basis of the use cases. Again, it depends on how many parts of a particular feature you require. There is a pricing calculator to get an aggregate cost based on the use cases, operating systems, storage capacity, etc. For Windows, the pricing depends on which side of the node you are using, i.e., Dual-core 4 GB RAM or Dual-core 8 GB RAM. It is possible to receive discounts on pricing with reserve incentives. In this case, it relies on how many VMs you run in the back end.

I advise others to initiate with basic services of the solution and then explore further. It is easy to maintain the solution. We can view and monitor all the processes through its monitoring feature. I rate the solution as a nine.

The AKS service is used as container orchestration for a product based on microservice architecture and as computing layer abstraction.

The solution helped the organization improve the availability and scalability of the product by making use of self-healing and autoscaling features offered.

The solution is a managed Kubernetes, allowing to offload of a large part of the maintenance tasks to the cloud service provider. This allows the team to focus more on features that have a greater business impact, rather than on maintenance tasks.

The solution provides a good foundation but there is a good amount of work to do to reach a full-fledged production grade platform.

There is an entire ecosystem of tooling around the solution. 

Th solution could be an improved by providing an add-on ready to install for the most common tooling around Kubernetes.

I have been using the solution for almost three years.

The platform is pretty mature.

The solution can easily adapt to the change of demand.

The initial setup has been very simplified and optimized compared to the effort required to deploy a self-hosted solution.

AKS provides a good level of elasticity allowing them to scale horizontally the platform based on demand.

I give the solution a nine out of ten.

Maintenance is much easier than an on-prem/self-hosted Kubernetes platform. 

Many of the tasks are automated by the cloud provider.

Our primary use case is to host our applications for Azure Cloud, rather than hosting the application on our on-premises infrastructure. We are hosting our applications in Kubernetes in Azure Cloud. I work for a product waste company, central components wherein during the weekend the traffic is more and during the weekdays the traffic is less. During festival time, the web traffic is more. So we need some kind of automatic solution. When the web traffic is more, the Kubernetes Service auto-scales, increasing the number of resources like horizontal scaling or vertical scaling. And if the web traffic is less, it descales, like releasing some of the resources. Kubernetes helps in autoscaling and that is why we migrated from on-premises to Kubernetes.

The most valuable feature is the autoscaling and self-healing. In our application, we have multiple replicas. If one of the replicas is not working due to anything, a new replica is automatically created.

I would like to see a graphical user interface.

I have been working with Azure Kubernetes Service (AKS) for the past three years.

There is excellent stability.

There is above-average scalability and we do have plans to increase our usage.

The initial setup is complex and not straightforward. I think it would be helpful if we had some type of script for the installation to make it easier.

We used a vendor team lens from the Mirantis company.

Minikube is one small Kubernetes cluster. With minikube, we can install this on our local laptop and we can practice. So to know Kubernetes, we have a concept of minikube which can be installed on a local laptop and we can practice Kubernetes commands. On a scale of one to ten, I would rate Azure Kubernetes Service (AKS) an eight.