Optimize the document of QOS

docs/tutorials/using-qos-ensurance.md

@@ -1,11 +1,11 @@
 # Qos Ensurance
 QoS ensurance guarantees the stability of the pods running on Kubernetes.
 
-It has the ability of interference detection and active avoidance. When pod with higher priority is affected by resource competition, disable schedule, throttle and evict will be applied to pod with lower priority to ensure the overall stability of the node. Currently, the absolute value/percentage of the node's cpu/mem is supported as a watermark. For details, please refer to qos-interference-detection-and-active-avoidance.md.
-When there is interference for eviction or throttle, accurate calculation will be performed, and the operation will be stopped when the load is lowered to slightly below the watermark to prevent accidental injury and transitional operation. For details, please refer to qos-accurately-perform-avoidance-actions.md.
+It has the ability of interference detection and active avoidance. When pod with higher priority is affected by resource competition, disable schedule, throttle and evict will be applied to pod with lower priority to ensure the overall stability of the node. Currently, the absolute value/percentage of the node's cpu/mem is supported as a watermark. For details, please refer to [Interference Detection and Active Avoidance](qos-interference-detection-and-active-avoidance.md).
+When there is interference for eviction or throttle, accurate calculation will be performed, and the operation will be stopped when the load is lowered to slightly below the watermark to prevent accidental injury and transitional operation. For details, please refer to [Accurately Perform Avoidance Actions](qos-accurately-perform-avoidance-actions.md).
 
-At the same time, Crane supports custom metrics to adapt to the entire interference detection framework. Users only need to complete some operations such as sorting definition, and can reuse the interference detection and avoidance processes including precise operations.For details, please refer to qos-customized-metrics-interference-detection-avoidance-and-sorting.md.
+At the same time, Crane supports custom metrics to adapt to the entire interference detection framework. Users only need to complete some operations such as sorting definition, and can reuse the interference detection and avoidance processes including precise operations.For details, please refer to [Define your watermark](qos-customized-metrics-interference-detection-avoidance-and-sorting.md).
 
-Crane has the dynamic resource oversold ability enhanced by the prediction algorithm, and reuses the idle resources. At the same time, it combines the prediction ability of the crane to better reuse the idle resources. Currently, idle resource recycling of cpu and mem is supported. At the same time, it has the elastic resource limitation function to limit the workload of reusing idle resources, and avoid impact on high-quality business and starvation issues. For details, please refer to qos-dynamic-resource-oversold-and-limit.md.
+Crane has the dynamic resource oversold ability enhanced by the prediction algorithm, and reuses the idle resources. At the same time, it combines the prediction ability of the crane to better reuse the idle resources. Currently, idle resource recycling of cpu and mem is supported. At the same time, it has the elastic resource limitation function to limit the workload of reusing idle resources, and avoid impact on high-quality business and starvation issues. For details, please refer to [Dynamic resource oversold and limit](qos-dynamic-resource-oversold-and-limit.md).
 
-At the same time, it has enhanced bypass cpuset management capability to improve resource utilization efficiency while binding cores. For details, please refer to qos-enhanced-bypass-cpuset-management.md.
+At the same time, it has enhanced bypass cpuset management capability to improve resource utilization efficiency while binding cores. For details, please refer to [Enhanced bypass cpuset management capability](qos-enhanced-bypass-cpuset-management.md).

docs/tutorials/using-qos-ensurance.zh.md

@@ -2,11 +2,11 @@
 Qos Ensurance 保证了运行在 Kubernetes 上的 Pod 的稳定性。
 
 具有干扰检测和主动回避能力，当较高优先级的 Pod 受到资源竞争的影响时，Disable Schedule、Throttle以及Evict 将应用于低优先级的 Pod，以保证节点整体的稳定，
-目前已经支持节点的cpu/mem负载绝对值/百分比作为水位线，具体可以参考qos-interference-detection-and-active-avoidance.zh.md
-在发生干扰进行驱逐或压制时，会进行精确计算，将负载降低到略低于水位线即停止操作，防止误伤和过渡操作，具体内容可以参照qos-accurately-perform-avoidance-actions.zh.md。
+目前已经支持节点的cpu/mem负载绝对值/百分比作为水位线，具体可以参考[干扰检测和主动回避](qos-interference-detection-and-active-avoidance.zh.md)
+在发生干扰进行驱逐或压制时，会进行精确计算，将负载降低到略低于水位线即停止操作，防止误伤和过渡操作，具体内容可以参照[精确执行回避动作](qos-accurately-perform-avoidance-actions.zh.md)。
 
-同时，crane支持自定义指标适配整个干扰检测框架，只需要完成排序定义等一些操作，即可复用包含精确操作在内的干扰检测和回避流程，具体内容可以参照qos-customized-metrics-interference-detection-avoidance-and-sorting.zh.md。
+同时，crane支持自定义指标适配整个干扰检测框架，只需要完成排序定义等一些操作，即可复用包含精确操作在内的干扰检测和回避流程，具体内容可以参照[定义自己的水位线指标](qos-customized-metrics-interference-detection-avoidance-and-sorting.zh.md)。
 
-具有预测算法增强的弹性资源超卖能力，将集群内的空闲资源复用起来，同时结合crane的预测能力，更好地复用闲置资源，当前已经支持cpu和mem的空闲资源回收。同时具有弹性资源限制功能，限制使用弹性资源的workload最大和最小资源使用量，避免对高优业务的影响和饥饿问题。具体内容可以参照qos-dynamic-resource-oversold-and-limit.zh.md。
+具有预测算法增强的弹性资源超卖能力，将集群内的空闲资源复用起来，同时结合crane的预测能力，更好地复用闲置资源，当前已经支持cpu和mem的空闲资源回收。同时具有弹性资源限制功能，限制使用弹性资源的workload最大和最小资源使用量，避免对高优业务的影响和饥饿问题。具体内容可以参照[弹性资源超卖和限制](qos-dynamic-resource-oversold-and-limit.zh.md)。
 
-同时具备增强的旁路cpuset管理能力，在绑核的同时提升资源利用效率，具体内容可以参照qos-enhanced-bypass-cpuset-management.zh.md。
+同时具备增强的旁路cpuset管理能力，在绑核的同时提升资源利用效率，具体内容可以参照[增强的旁路cpuset管理能力](qos-enhanced-bypass-cpuset-management.zh.md)。