Hi, Here is the summary of another proposal regarding job cluster management a few of us have been discussing.
By design, Heron topologies are totally independent from each other for isolation and maintainability. However, as the result, the deployment (and other topology management tasks) could be slow which is another operational burden. The proposal is about how to improve topology and job cluster management to ease the burden. Any thoughts could be super helpful to make the proposal better since there are different use cases. Please definitely feel free to reply. Thanks in advance! https://docs.google.com/document/d/1Oq8saQzrcJmUBaFYlVhY4HyevTOrkqzFZryAyVrkaFM/edit Regards, --ning