Apache Phoenix on Spark - Unable to insert to Phoenix HBase Tables/Need suggestion for Best practice -

i have table structure below.

trans_count

start_time,           end_time,                count  00:00:01              00:00:10                 1000  00:00:11              00:00:20                 800 

spark listens events kafka , grouping 10 seconds , have insert phoenix hbase table.

after 10 seconds, have first check if start_time,end_time combination in table. if there, have take existing count , add new count , upsert again.

upsert trans_count(start_time, end_time, count) select start_time, end_time, count? trans_count start_time = ? , end_time = ? 

if no rows upserted in above statement, upsert data.

in apache storm, able create phoenix connection object in configure method , able use same connection once every 10 seconds upsert.

in spark, not create connection object , use same object every object in rdd. output spark javadstream> start_time, end_time, count keys in map.

i end creating connection object every iteration of rdd, feel not right way. have read phoenix connections light weight creating connection object every rdd seems not right way.

i read blogs related stuff, not getting going. kindly help.

note : appliation built in java.

solution:

instead of creating connection per objectin rdd, create connection per partition in rdd , use objects.