I’m looking at the very nice MapReduce implementation in Ignite and have a question about the reduce stage.
It appears that the reduce stage only occurs once all mapped results have been accumulated. I have a use case where the size of that data set is very large and problematic to hold in memory until all mapped results are returned. Is it possible to do a continuous Reduce in Ignite, similar to the way it is possible to do a continuous Map? I can see one possible way of doing this by slurping off the responses as they are provided to OnResult, but I’m wary of mutating the list provided to that method to pull mapped results out of the list being accumulated. Thanks, Raymond.
