yinzhijian edited a comment on pull request #8553: URL: https://github.com/apache/incubator-doris/pull/8553#issuecomment-1074864940
> I personally think that the scalability of this implementation is not good, especially when there are a large number of machines. If there is data skew in the internal be for each machine, then the repair bottleneck will be thrown to fe. now cluster balancing is to deal with data skew between bes. 1. Personally, I think that disk balancing does not affect the reliability of data like recovery, so its priority is not high. If it is completed too quickly, it will affect the performance of BE, thereby affecting service stability. >Then this implementation and cluster balance can run together？how do they collaborate with each other? when the tablet is scheduled to a be, then the be also needs to consider the disk strategy 2. By default, disk balancing and cluster balancing will not run together. Only when the cluster has been balanced, the disk balancing task will be scheduled. Thereby avoiding unnecessary data migration. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
