Is it possible to implement hierarchical timing wheels solution: http://www.cs.columbia.edu/~nahum/w6998/papers/sosp87-timing-wheels.pdf True that your proposal is relatively easy to implement, but it introduces too many random reads. Operating systems normally read data in pages(at least 4k) with potential read-ahead, making page cache pollution more serious.
The hierarchical timing wheels solution, however, is capable of making sequential reads/writes, thus much more performant. As a matter of fact, Aache Kafka has already implemented this solution: https://www.confluent.io/blog/apache-kafka-purgatory-hierarchical-timing-wheels/ According to their description, the engineer complexity is contained and worthwhile. On Tue, Jun 21, 2022 at 12:01 PM yukon <[email protected]> wrote: > The long-awaited feature for our community, this RIP will make our timer > message more competitive, and make the new SimpleConsumer API > `changeInvisibleDuration` more flexible. > > On Tue, Jun 21, 2022 at 11:42 AM dongeforever <[email protected]> > wrote: > > > Good Job. > > > > It will be nice if there are some performance test reports. > > > > 季俊涛 <[email protected]> 于2022年6月20日周一 17:28写道： > > > > > Hi, RocketMQ Community: > > > > > > Currently, RocketMQ's delay message feature only supports delayed > > delivery > > > for specific time levels. Such delay message feature(only support > > specific > > > levels of delay time) is no longer enough to support the flexible usage > > of > > > rocketmq. Therefore, we need a delay message feature that supports > > > arbitrary delay time. So I would like to start an email thread to > discuss > > > RIP-42 Support Timing Messages with Arbitrary Time Delay. > > > > > > I have written my proposal and you can click on the link below： > > > Google Doc: > > > > > > https://docs.google.com/document/d/1D6XWwY39p531c2aVi5HQll9iwzTUNT1haUFHqMoRkT0/edit?usp=sharing > > > Shimo:https://shimo.im/docs/gXqme9PKKpIeD7qo/ > > > > > > If you have any questions or suggestions, please reply to this email or > > > comment on the proposal. > > > > > > Thanks > > > Juntao Ji > > > > > > > > >
