Azure Kubernetes Service (AKS) vs. Google Cloud Spanner

AKS works very well for running containerized applications that require high availability and scalability. This includes systems like our HRIS platform and customer-facing web applications. AKS is a good choice when applications are broken into multiple services that need independent scaling and deployment. It provides the flexibility needed to manage these architectures effectively. But for single, low-traffic applications or simple internal tools, AKS can be overkill. For scenarios like that Azure App Service would be better.

Incentivized

Google Cloud Spanner is suited for limitless horizontal scaling while maintaining strong consistency which needs to support ACID. NoSQL databases work in scaling but no ACID support. RDBMS support ACID, but horizontal scaling is not as great. The API it provides result in some limitations to related areas of the code, such as connection pools or database linking framework. So high # of connection pools can vary.

Incentivized

• AKS makes it easier to replicate data to multiple regions
• Azure portal make it easier to manage the resources of the organization

Incentivized

• Super high availability
• Scales automatically
• High standard SLA

Incentivized

• Steep learning curve
• Expected charges are unclear until you see real production usage
• Operations teams need to learn an entirely new skill set

Incentivized

• Support for Views
• Support for more databases (schemas).
• More index types that can be supported (Functional)
• Backups (ie table/data backup) if data is deleted or truncate by accident.

Incentivized

As already said, the UI/CLI and even terraform are perfectly fine, but certain details could be documented better. For instance, if I want to secure the whole Azure Kubernetes Service (AKS) with my own managed keys, then it is very complex and hard to get there. Not really a single source that gives you the whole picture. Besides that, it is still good to use, in most cases intuitive but details mentioned as above can be tricky.

Incentivized

Microsoft support was really good, whenever we raise any ticket they come back to us within a couple of hours.

Incentivized

Amazon EKS stacked up very well and had better performance in some areas. However, Azure Kubernetes Service was a better fit given our Azure environment.

Incentivized

At that point, we were looking at something [that] can hold our relational database, [...] provide stable connection, and maintain high ACID transition. BigTable is for nonrelational database so it was out of our [sight] very quickly. BigQuery is a data warehouse that can hold huge amount of data but not ideal for transition. AWS RDS is [...] similar to Spanner but because most of our services are already on GCP, so we went with Spanner.

Incentivized

• We had to spend more time on Azure Kubernetes Service (AKS) than on AWS and GCP to get our kubernetes cluster up and running
• The resources on nodes need to be left out unused, so effectively it is wasting money there
• It definitely made us spend more time into maintaining kubernetes

Incentivized

• Backups specifically if transactional data is deleted. Restoring made us lose time.
• Sharding on Horizontal level was quick and easy. Deployment and increasing nodes is easy
• Large dataset handling.
• ACID compliance

Incentivized