Introduction

Azure Kubernetes Service (AKS) provides a seamless integrated experience for deploying and managing containerized applications in the cloud using Kubernetes. One of the critical elements of deploying applications in AKS is understanding the architecture of the underlying node disks. Azure offers several disk types for use with AKS, each catering to different performance and scalability needs.

In this article, we'll delve into the details of Azure AKS cluster node disk types, exploring their characteristics, use cases, and considerations for selection.

Node Disk Types in AKS

Azure AKS nodes use Azure Virtual Machines (VMs), which have their associated storage. The main disk types available are:

1. OS Disk
2. Temporary Disk
3. Data Disk

OS Disk

Purpose and Functionality:

• Used to store the operating system and boot volume.
• Persisted storage, meaning the data is retained even if the VM is deallocated or stopped.

Characteristics:

• Size: The size can be customized depending on the VM size and requirements (up to 2 TB).
• Performance: Managed Disks offer different tiers like Standard HDD, Standard SSD, and Premium SSD, each providing different throughput and IOPS.
• Recommendations: Choose SSD for faster boot times and overall system responsiveness, especially in production environments.

Temporary Disk

Purpose and Functionality:

• Provides temporary storage directly attached to the VM.
• Data is lost on VM deallocation or stoppage, hence ideally suited for ephemeral workloads.

Characteristics:

• Usage: Commonly used for swap space or temporary data processing.
• Performance: Exceptional performance because it's physically attached to the VM, with high IOPS and low latency.

Recommendations: Do not store critical data on temporary disks. They are best used for caching or other non-critical temporary storage purposes.

Data Disk

Purpose and Functionality:

• Attached to VMs to provide additional persistent storage.
• Useful for application data, logs, or setup that needs to persist independently of the OS.

Characteristics:

• Size: Support up to 32 TiB per disk.
• Performance: Similarly to OS disks, data disks also come in different performance tiers - Standard HDD, Standard SSD, and Premium SSD, and Ultra Disk (offering extreme performance).
• Scale: Multiple data disks can be attached to a VM, increasing total storage availability and performance via striping.

Recommendations: Data disks are appropriate for applications requiring high availability and reliability. Choose based on the workload's IOPS and throughput requirements.

Table of Disk Types

Here is a summary table showing key characteristics for the different node disk types:

Additional Considerations

Cost Considerations

• Overall Costs: The costs associated with each disk type vary based on size, performance tier, and region.
• Performance vs. Cost Balance: It's crucial to balance performance needs against budget constraints. Premium SSDs offer superior performance but at a higher cost compared to standard options.

Scalability and Flexibility

• Dynamic Resizing: Both OS and data disks can be resized as necessary, providing flexibility in scaling resources up or down based on application demands.
• Attachment Limits: Be aware of maximum disk attachment limits per VM size when planning capacity.

Performance Metrics

• Latency and IOPS: The right choice of disk type can significantly impact application latency and IOPS. Ultra Disks provide the highest IOPS, suitable for intensive applications requiring sub-millisecond latencies.
• Disk Throughput: Evaluate workloads and choose higher-throughput disk options for IO-intensive applications.

Conclusion

Selecting the appropriate disk type in Azure AKS is crucial for achieving a balance between performance, cost, and resilience. While the temporary disk is suitable for ephemeral data needs, the OS and data disks provide durable storage options tailored to different workload requirements. Understanding each disk's characteristics allows for informed decision-making, leading to optimally designed and effective Kubernetes clusters in Azure.