what situation does index works?

[xuxc]

environment：
hadoop-2.3.0-cdh-5.1.0
hbase-0.98.1-cdh-5.1.0
phoenix-4.1.0

i first create table in phoenix-client (or create it in hbase shell,and then do it again in phoenix client),and then create index on that table;
if I use "upsert" clause to insert data,the record corresponding to index_table is useful;
but when I use ETL tool(kettle spoon) to load data to Hbase,i found the corresponding index table is empty.
For 10million rocords,one query take almost 50 sec because index does not works.
i want to if any mistakes i do that.
thx.

[jtaylor-sfdc]

Please subscribe and post your question on our current email Apache
address: http://phoenix.apache.org/mailing_list.html. Phoenix moved to
Apache about a year ago and is a top level project now. This site is
defunct now.