𝗖𝗣𝗨 𝗧𝗵𝗿𝗼𝘁𝘁𝗹𝗶𝗻𝗴 𝗶𝗻 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 ☸️

𝗖𝗣𝗨 𝗧𝗵𝗿𝗼𝘁𝘁𝗹𝗶𝗻𝗴 𝗶𝗻 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 ☸️ : occurs when a container’s CPU usage hits its limit, causing Kubernetes to restrict CPU access, leading to potential performance degradation.
𝗛𝗼𝘄 𝗖𝗣𝗨 𝗧𝗵𝗿𝗼𝘁𝘁𝗹𝗶𝗻𝗴 𝗪𝗼𝗿𝗸𝘀 𝗶𝗻 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 ☸️
In Kubernetes, CPU resources for containers are managed with CPU requests and CPU limits:

𝟭. 𝗖𝗣𝗨 𝗥𝗲𝗾𝘂𝗲𝘀𝘁𝘀: This is the minimum amount of CPU that Kubernetes guarantees to a container. When a container requests a certain CPU amount, Kubernetes tries to ensure that amount is available on the node.

𝟮. 𝗖𝗣𝗨 𝗟𝗶𝗺𝗶𝘁𝘀: This is the maximum CPU that a container is allowed to use. If a container tries to exceed this limit, Kubernetes throttles the CPU usage by enforcing the limit.

Kubernetes uses the Completely Fair Scheduler (CFS) to enforce CPU limits. When a container exceeds its limit, CFS reduces its CPU usage by briefly "pausing" it, impacting performance and increasing latency, especially for high-demand tasks like initialisation or intensive processing.

𝗞𝗲𝘆 𝗿𝗲𝗮𝘀𝗼𝗻𝘀 𝗳𝗼𝗿 𝗖𝗣𝗨 𝘁𝗵𝗿𝗼𝘁𝘁𝗹𝗶𝗻𝗴 𝗶𝗻𝗰𝗹𝘂𝗱𝗲:
• 𝗧𝗶𝗴𝗵𝘁 𝗖𝗣𝗨 𝗟𝗶𝗺𝗶𝘁𝘀: If CPU limits are set too low relative to the container’s demand, the container quickly hits the limit and gets throttled.
• 𝗛𝗶𝗴𝗵 𝗖𝗣𝗨 𝗦𝗽𝗶𝗸𝗲𝘀: Background tasks like garbage collection (GC) or intensive computations can trigger high CPU usage that exceeds limits.
• 𝗟𝗶𝗺𝗶𝘁𝗲𝗱 𝗖𝗹𝘂𝘀𝘁𝗲𝗿 𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀: In environments with high contention, such as dev or CI environments with strict limits, containers are more likely to hit CPU limits.

𝗜𝗻𝗱𝗶𝗰𝗮𝘁𝗼𝗿𝘀 𝗼𝗳 𝗖𝗣𝗨 𝗧𝗵𝗿𝗼𝘁𝘁𝗹𝗶𝗻𝗴:
• 𝗰𝗼𝗻𝘁𝗮𝗶𝗻𝗲𝗿𝗰𝗽𝘂𝗰𝗳𝘀𝘁𝗵𝗿𝗼𝘁𝘁𝗹𝗲𝗱𝘀𝗲𝗰𝗼𝗻𝗱𝘀_𝘁𝗼𝘁𝗮𝗹: This metric shows the cumulative time a container has been throttled, providing insight into the frequency and duration of throttling.
• 𝗣𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗗𝗲𝗴𝗿𝗮𝗱𝗮𝘁𝗶𝗼𝗻: Symptoms like increased response times, slower processing, or latency spikes can suggest throttling is affecting the application.

𝗧𝗼 𝗿𝗲𝗱𝘂𝗰𝗲 𝗖𝗣𝗨 𝘁𝗵𝗿𝗼𝘁𝘁𝗹𝗶𝗻𝗴, 𝘆𝗼𝘂 𝗰𝗮𝗻:

• 𝗔𝗱𝗷𝘂𝘀𝘁 𝗖𝗣𝗨 𝗟𝗶𝗺𝗶𝘁𝘀: Ensure limits are realistic for the container’s expected workload.
• 𝗦𝗲𝘁 𝗖𝗣𝗨 𝗥𝗲𝗾𝘂𝗲𝘀𝘁𝘀 𝗖𝗮𝗿𝗲𝗳𝘂𝗹𝗹𝘆: Balancing requests and limits prevents frequent throttling while reserving sufficient CPU.
• 𝗠𝗼𝗻𝗶𝘁𝗼𝗿 𝗮𝗻𝗱 𝗣𝗿𝗼𝗳𝗶𝗹𝗲 𝗨𝘀𝗮𝗴𝗲: Tracking CPU usage over time can reveal the need for higher limits or more efficient resource allocation.
• 𝗜𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁 𝗔𝘂𝘁𝗼𝘀𝗰𝗮𝗹𝗶𝗻𝗴: For high-demand periods, horizontal autoscaling can provide more resources to avoid throttling.
By balancing CPU allocation and setting realistic limits, CPU throttling can be minimised, enhancing performance and stability for applications in Kubernetes