Open Rstar1998 opened 1 year ago
No an expert on Beam's support on this. But have you tried the debezium io? https://beam.apache.org/releases/javadoc/2.30.0/org/apache/beam/io/debezium/DebeziumIO.html?
@xinbinhuang You are right. But a managed service (wrt to compute and scaling) is always preferred by most of the customers rather than hosting debezium on a server. If we are able to make mongo streaming possible using dataflow , it will help in lot of data engineering usecases
I'm not sure if I follow your questions. I believe the DebeziumIO
is self-contained and encapsulated within Beam, so you don't need to set up a separate kafka and debezium cluster. Though experimental, it should also work with the MongoDbConnector
.
If we are able to make mongo streaming possible using dataflow , it will help in lot of data engineering usecases
Is your use case about Beam or about dataflow? Beam should just work on dataflow.
@xinbinhuang DebeziumIO should work in my case. But unfortunately it doesn't have Mongo support in beam library itself.
https://beam.apache.org/releases/javadoc/2.30.0/org/apache/beam/io/debezium/DebeziumIO.html
Maybe some tweaks will able to do our job. Because debezium does have Mongo connector support.
@benWize is it possible to include mongo connector in debezium section of apache beam ?
What would you like to happen?
Hi folks , There are many use cases where people want realtime migration of collections of Mongodb to Big query (Data Warehouse ) / GCS (Data lake). Hence , it would be great if apache beam mongo sdk included support to read Mongodb cdc (Change data capture) streams in real time and those events capture can be dumped into big query or GCS for analytical purposes. Currently to achieve this stuff we need a lot complex architecture in GCP including debezium , kafka etc. If this feature is implemented properly there will be only one single pipeline which will do all the stuff.
Issue Priority
Priority: 3
Issue Component
Component: io-py-mongodb