json - Stream from JMS queue and store in Hive/MySQL -
i have following setup (that cannot change) , i'd advice people have been down road. i'm not sure if right place ask, here goes anyway.
- various json messages placed on different channels of jms queue (universal messaging/webmethods).
- before data can stored in relational-style dbs has transformed: renamed, arrays flattened , structures nested objects extracted.
- data has appended mysql (as serving layer visualization tool) , hive (for long-term storage).
- we're stuck on spark 1.4.1 , may move 1.6.0 in few months' time. so, structured streaming not (yet) option.
- at point events streamed directly real-time dashboards, having in place capable of doing ideal.
- ideally coding done in scala (because have considerable batch-based repo spark , scala), minimal requirement jvm-based.
i've looked @ spark streaming not have jms adapter , far can tell operating on json done using sqlcontext instance on dstream's rdds. understand it's possible write custom adapter, i'm not sure if spark still best/easiest solution. i've looked @ doc samza , flink did not find jms and/or json, @ least not natively.
apache camel seems might have substantial set of connectors i'm not familiar it, , impression not streaming part, 'just' bit connect various systems. there's akka although impression it's more of replacement messaging systems , jms set.
there bewildering amount of available tools , i'm @ point @ loss @ or out for. recommend based on experience use pick messages, transform, , insert hive , mysql?
Comments
Post a Comment