Apache Kudu

What do you like best?

1. Implementing Lambda Architecture (Both Batch & Real Time Streaming) with Kudu is quite straightforward. We additionally used Streamsets as the Ingestion Platform which has quite a good coupling with Kudu.

2. Makes Real Time Analytics Quite Starightforward. We used Kudu to run multiple Real Time Campaigns.

3. Tailor Made for Implementing Data Warehouses within a Big Data Environment.

4. Nice Upsert Functionality.

What do you dislike?

1. Partition Limitation - Is limited to 2000 tablets/tablet server.

2. Faces Random Timeouts while approaching the tablet number limitation (max 2000/server).

3. Needs a Conformed Schema. No automatic handling of drifting schemas.

4. Needs a Primary Key for every table.

5. #3 & #4 are not really cons per se.

Recommendations to others considering the product:

Use it if the Datawarehouse is a strong use case within Big Data enviroments & try to be within the partition/tablet limitation and you should be good to go.

What problems are you solving with the product? What benefits have you realized?

1. Real Time Campaigns.

2. Real Time Lookups & Transaction Enrichment.

3. Data Warehouse Implementation is quite straight forward with Apache Kudu.