You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
What would you like to be added:
Currently, the forward plugin offers only two static load balancing algorithms: round robin and sequential. I propose the integration of dynamic algorithms, such as load balancing based on the response time of upstream DNS servers. Why is this needed:
In my setup, I've configured three upstream DNS servers in the forward plugin, utilizing the sequential load balancing algorithm. However, when the latency of upstream server A increases, it impacts the entire Kubernetes cluster by elongating the time taken for name resolution when establishing new connections to certain hosts. I advocate for the inclusion of dynamic load balancing algorithms within the forward plugin to address this issue. Utilizing metrics like ResponseTime - QueryTime or something akin to the coredns_forward_request_duration_seconds metric could facilitate the implementation of a dynamic load balancing strategy. This would mitigate the adverse effects caused by latency spikes in individual upstream servers, ensuring smoother operation of the Kubernetes cluster.
The text was updated successfully, but these errors were encountered:
What would you like to be added:
Currently, the forward plugin offers only two static load balancing algorithms: round robin and sequential. I propose the integration of dynamic algorithms, such as load balancing based on the response time of upstream DNS servers.
Why is this needed:
In my setup, I've configured three upstream DNS servers in the forward plugin, utilizing the sequential load balancing algorithm. However, when the latency of upstream server A increases, it impacts the entire Kubernetes cluster by elongating the time taken for name resolution when establishing new connections to certain hosts. I advocate for the inclusion of dynamic load balancing algorithms within the forward plugin to address this issue. Utilizing metrics like
ResponseTime - QueryTime
or something akin to thecoredns_forward_request_duration_seconds
metric could facilitate the implementation of a dynamic load balancing strategy. This would mitigate the adverse effects caused by latency spikes in individual upstream servers, ensuring smoother operation of the Kubernetes cluster.The text was updated successfully, but these errors were encountered: