Overall Satisfaction with IBM Cloud Kubernetes Service
We are migrating our current application from IBM Cloud Foundry to Kubernetes. We grew out of CF and needed more compute power, therefore we turned to Kubernetes. Currently, we are using it as the primary architecture for our UI and a few of our complex processing processes (pseudo batch processing). We are implementing it for this one client, but we have other clients on Kubernetes on other cloud platforms. The application that it is running supports 300 million dollars in business for this company. The solution is only for a single subset of a larger department. Many pieces of the application are still in Cloud Foundry; we are still in the process of migrating.
- It scales really well
- Depending on how you measure, it is cheaper and more powerful than Cloud Foundry
- Deployments are fast, comparably.
- It is very unreliable. We have seen nodes and machines randomly go down without warning or helpful error messages.
- Documentation is not great.
- The autoscale and documentation on autoscale are not good. We had to write/re-write our own and as a result; we found a lot of issues that caused problems.
- Batch processes that used to take hours, we brought down to minutes or an hour.
- We have found it to be more buggy or less stable than other cloud offerings.
- Lack of documentation has been challenging.
- Lack of positive experience might lead us to recommend a different cloud provider.
- Google Kubernetes Engine, IBM Cloud Foundry and Amazon Elastic Compute Cloud (EC2)
Internally, we have evaluated and used Google Kubernetes Engine, Azure Kubernetes, and AWS Kubernetes. Overall, our team would rank IBM's offering as 2nd or 3rd amongst those we have used and deployed thus far. We selected IBM Kubernetes here because the client has almost all of their assets with IBM. K8 is better than Cloud Foundry for many things. The scaling is much better than running things on lots of EC2 instances.