Google Cloud Run vs. Google Kubernetes Engine

Microservices and RestFul API application as it is fast and reliant. Seamless integration with event triggers such as pubsub or event arc, so you can easily integrate that with usecases with file uploads, database changes, etc. Basically great with short-lived tasks, if however, you have long-running processses, Cloud Run might not be idle for this. For example if you have a long running data processing task, other solutions such as kubeflow pipelines or dataflow are more suited for this kind of tasks. Cloud Run is also stateless, so if you need memory, you will have to connect an external database.

Incentivized

Google Kubernetes Engine is well suited for dynamic and large workloads since it can scale up with usage. It is easily configurable, which allows for flexibility. User interface is simple to navigate, which reduces roadblocks for a team with people unfamiliar with Kubernetes. Great if you are already using other GCP services as it integrates well with that.

Incentivized

• Real-time autoscaling. Escalamento automático em tempo real
• Simplified Continuous Deployment. Implantação contínua simplificada
• Running tasks in the background. Execução de tarefas em segundo plano

Incentivized

• Deployment of a new GKE cluster is really fast in comparison to other cloud providers.
• GCP is ahead other vendors and always provide the most up to date Kubernetes version.
• GKE automation for master upgrade and the worker nodes pool works really well.

Incentivized

• Cloud Run doesn't allow you to redeploy an already existing revision which can be inconvenient in some use cases
• Tricky to get the deployment working to start but once it's working that's great
• The actual deployment is not the fastest but it's not too bad

Incentivized

• Not as intuitive as it could be
• Documentation could be better, especially for people using other Google Cloud tools
• Not the preferred Kubernetes Engine for many apps

Incentivized

We definitely need to renew it because we dont own our own infrastructure and storage and we are happy with Cloud Run features

Incentivized

The UI/console is great... the documentation is top-notch for developers, but the CLI itself when you have to script around it is very complex and easy to forget some options... the downside of a generic command line client.

Incentivized

It's a great product if you learn it. It has flexibility and is very strong. Autoscaling and Resource management make running huge applications a breeze. Using Helm with Kubernetes and Terraform for infrastructure creation can totally automate your CICD pipeline. You also get easy access to CUDA cores for machine learning.

Incentivized

Not seen any major issues when we run applications its good

Incentivized

Initially we felt slow but slowly it picked up and easy to manage

Incentivized

Google support is excellent and helpful, but the first answer is always so bureaucratic no matter how many logs, evidence, and information you sent.

Incentivized

I was involved in the initial implementation setup, Its easy with the given documentaiton we can do ourself. Not that critical

Incentivized

AWS Lambda supports code zip package, enabling lower cold start time. Also, AWS Lambda pricing is much simpler, easier to understand.

Other than that, the 2 products are very similar, including the Docker image support: the image must be built based on proprietary base image.

Obviously, if your other services are running in GCP, then Google Cloud Run is your only choice for tight integration, & private networking.

Incentivized

We had to move several products to Google Cloud, and the Google Kubernetes Engine was the option recommended to us, so we investigated it and ran with it. Back then (2019), we were not aware of Cloud Run-provisioned K8s clusters, so our other option was a completely self-managed K8s cluster on Compute Engine VMs, which we did not have the knowledge of and capacity to handle.

Incentivized

It has good auto scale feature and reliable also

Incentivized

• It has saved us some costs since we now do not require a live server and have moved to a serverless workflow for these services
• Breaking changes do not affect the entire application now that we have separated our concerns using a serverless service
• Much easier to debug since we can now isolate our services and reduce the search space for finding/fixing bugs

Incentivized

• Compared to other big K8s providers it has the best price/performance factors.
• Upgrade process from stable to regular versions
• Old stable releases: 1.15/1.16 should be in a stable branch.

Incentivized