For now, dubbo-admin makes use of proven components to build a user-friendly admin interface to help users manage and monitor microservice-oriented systems. But there is still gap from excellent mesh control plane like istio. There is no definitive conclusion as to which of the SDK model and sidecar model is better. In order to compete and improve the dubbo ecology, we should benchmark istio to improve the control plane function and play our own advantages.
I propose to accomplish this goal from the following two points.
dubboctl
pr: https://github.com/apache/dubbo-admin/pull/1025. At the first phase, generate and install sub-command are the most important part. Based on this phase, dubboctl could evolve into operator. In the process, more and more commands about monitoring and managing are being integrated like istioctl.
control plane abstraction layer
Implement most of functionalities of mesh control plane by ourselves is not cost-effective. We should build a high-level abstracttion layer to integrate as many existing proven components as possible and provide users more convenience and extensibility. The details will be presented after a detailed study.
Mesh Control Plane Proposal
For now, dubbo-admin makes use of proven components to build a user-friendly admin interface to help users manage and monitor microservice-oriented systems. But there is still gap from excellent mesh control plane like istio. There is no definitive conclusion as to which of the SDK model and sidecar model is better. In order to compete and improve the dubbo ecology, we should benchmark istio to improve the control plane function and play our own advantages.
I propose to accomplish this goal from the following two points.