Besides the fact that it will add overhead to your writes depending on what streaming guarantees you desire (at-most-once, at-least-once, exactly-once), how many subscribers you want to support, the restart capabilities you would want in failure scenarios, etc.
Rohit On Oct 24, 2017, at 11:44 PM, Josh Elser <[email protected]<mailto:[email protected]>> wrote: The most reliably way (read-as, likely to continue working across HBase releases) would probably be to implement a custom ReplicationEndpoint. This would abstract away the logic behind "tail'ing of WALs" and give you some nicer APIs to leverage. Beware that this would still be a rather significant undertaking that would likely require you to dig into HBase internals to get correct. On 10/24/17 4:02 PM, yeshwanth kumar wrote: Hi i am searching for a way to stream data from hbase, one way to do is with filters , but i need to query hbase continously, another way is to read directly from WAL, (i am searching for sample code, and i found WALReader and WAL.Entry API's. can i use them directly without any side effects) can anyone suggest me a good way to stream data out of hbase, as the write happens, i want the same data to be pushed to another data source. please let me know -Yeshwanth Can you Imagine what I would do if I could do all I can - Art of War
