Using the Phoenix client to query data error

Hello, The hoenix is ​​the client implementation of HBase, hoenix has HBase features, HBase is good at making Phoenix, and HBase is not good at Phoenix. For this kind of aggregate function like count/groub by, Phoenix Processing is also very weak, the implementation is not fast, it is easy to time out, the timeout is divided into two layers, one is the query timeout of Phoenix itself, one is the timeout of the scan after the magazine is disassembled into scan, see from the above log, this Obviously the latter, socket timeout.

Divide your problem into two resolution levels:

1. Analyze whether the statement is reasonable to use in Phoenix, and whether it can meet the requirements in other ways, such as MR tasks.

2. If the client has to solve with SQL, then analyze the table and statement to see if you can reduce the timeout by changing the table structure, such as placing the query column into a separate column family.

3. Even if the tables and statements are analyzed and the table structure and statements are optimized, there may still be timeouts, so the two timeout parameters of scan are moderately large.