Hi,
Here is what I understood from what I read but from implementation I have a few questions.
Please let me know if I can configure my system like the below.
1 server running serving layer
2 or 1 servers running speed layer , batch , kafka and spark all together
x nodes of hadoop.
Streaming data comes to kafka and speed layer using spark builds the model updates kafka topic
For batch I feel data has to come from some etl tool storing to hadoop hdfs and batch builds the model using spark stores into kafka topic.
Serving layer reads kafka topic and serves the request via rest api from external clients request be it mobile.
Please help if the above servers 1,2 and 3 are feasible options.
Hi, Here is what I understood from what I read but from implementation I have a few questions. Please let me know if I can configure my system like the below.
Thanks