Hi guys I’m working on building lambda architecture of fully distribution in hadoop
I decide to use the following systems in each layers
Serving: kafka cassandra Batch: hadoop spark Speed: Storm
The batch layer now is ok in the multi machines while I’m confused with the others installations
I find that the remaining systems look like no master and slave architecture. Should I just install serving and speed layers in the master node of hadoop? Or they should be installed as well as hadoop or spark in all master and slave node? Or they should be setup in a independent machine?