Toivo, I looked into Akka very briefly several years ago and didn't think it was a good fit. But a lot has changed in the past several years.
It would likely be very useful to implement the Event-Driven scheduling strategy (which didn't even exist when I looked at Akka). Right now, that strategy is considered experimental. It actually performs worse than the Timer-Driven strategy and may have some issues at high rates. It was something that I started working on as a bit of an experiment and it could definitely use some work. A complete rewrite using Akka may make sense. It wouldn't be able to replace our scheduling model completely because we need to support the notion of CRON-based scheduling and timer-driven scheduling, but if we had the Event-Driven scheduling working well that would be a big plus. It's a good idea - feel free to experiment & ask questions :) ---------------------------------------- > Date: Sat, 20 Jun 2015 06:24:29 -0700 > From: [email protected] > To: [email protected] > Subject: Actor model and Akka > > I am curious. > Have someone considered using Akka internally for scheduling, clustering, > etc. > Nifi use internally heavily java.util.concurrent classes. > Nifi is carefully designed and implemented so concurrency problems are > solved. > Still this requires a lot of expertise and it's very easy to accidentally > introduce some hard to catch bugs. > > Actor model offers relatively easy way to handle concurrency. > Also Akka team have done lot of work in clustering area. > So maybe NiFi can reuse something? > > I have some Akka experience and using Actors is a lot safer than using > java.util.concurrent classes. > But Akka itself is too low level to built business applications directly on > top of Akka. > > Thanks > Toivo > > > > > -- > View this message in context: > http://apache-nifi-incubating-developer-list.39713.n7.nabble.com/Actor-model-and-Akka-tp1992.html > Sent from the Apache NiFi (incubating) Developer List mailing list archive at > Nabble.com.
