Thanks Guozhang. So there's no way we could also use InternalTopicManager to specify the number of partitions and RF?
https://github.com/apache/kafka/blob/0.10.1/streams/src/main/java/org/apache/kafka/streams/processor/internals/InternalTopicManager.java On 4 October 2016 at 19:25, Guozhang Wang <wangg...@gmail.com> wrote: > Hello Gary, > > What you described should be workable with the lower-level Processor > interface of Kafka Streams, i.e. dynamic aggregations based on the input > data indicating changes to the JSON schemas. For detailed examples of how > the Processor API works please read the corresponding sections on the web > docs: > > http://docs.confluent.io/3.0.1/streams/developer-guide.html#processor-api > > > Guozhang > > On Mon, Oct 3, 2016 at 6:51 AM, Gary Ogden <gog...@gmail.com> wrote: > > > I have a use case, and I'm wondering if it's possible to do this with > > Kafka. > > > > Let's say we will have customers that will be uploading JSON to our > system, > > but that JSON layout will be different between each customer. They are > able > > to define the schema of the JSON being uploaded. > > > > They will then be able to define the fields in that JSON they want to > > gather metrics on (sum, counts etc). > > > > Is there a way with Kafka streaming to dynamically read the configuration > > for that customer and process the json and do counts and sums for the > > fields they've defined. > > > > It's possible at any time they may want to modify the configuration for > > their json as well. Stop counting one field, start counting another. > > > > They will also want to do some inferences as well. IE, if this particular > > JSON is uploaded with a field in it, then check to see if another json > was > > uploaded within 8 hours. > > > > Is it possible for Kafka streaming to be this dynamic? > > > > > > -- > -- Guozhang >