Yes that is something that we are working on, especially around the resource aware scheduler. We have added in a new metrics path to have CPU and Memory usage reported to nimbus. We plan to extend this to other metrics in the future and then start working on how to properly do elasticity.
Thanks, Bobby On Sun, Feb 18, 2018 at 11:52 PM Ali Nazemian <[email protected]> wrote: > Hi All, > > Is there any plan to add autoscaling to Apache Storm? We are using Storm in > production heavily. However, one of the biggest challenges we have faced is > the lack of auto-scaling in Storm. Most of our Storm clusters are very > under-utilised because Storm relies on saving a slot for topologies > regardless of what the current load is. The Resource-Aware scheduler > may help us a little, but it doesn't scale well from the operational > perspective if you have tons of topologies to tune, so it is not what > exactly we need. What would be really nice to have is an auto-scaling > feature similar to Storm-594. Is there any plan to introduce this > functionality in future? > > Regards, > Ali >
