thank you for your reply.
nowadays,it is just a test cluster wtih 3 datanodes, data will also increase more in the future. the problem is: Any col (total 100+ cols) may become (also maybe not)the condition to set in "where" clause,and all cols must be get. and i want to get results within 3 sec for more than massive dataset. i have try to create index for every column,but when i conduct "select * "query ,the indexes don't work, so i want to know if there have any other ways? thanks. ------------------ 原始邮件 ------------------ 发件人: "Krishna";<research...@gmail.com>; 发送时间: 2014年10月25日(星期六) 凌晨2:19 收件人: "user@phoenix.apache.org"<user@phoenix.apache.org>; 主题: Re: for such condition,how to create index in phoenix? How many nodes are in the cluster? How fast do you expect this table to grow? In the current state, 1.5 mill rows is not a massive dataset, so, having an index for every possibility will result in more effort towards index maintenance and not produce comparable query improvements. If you can narrow down your use case to prioritize few of the column qualifiers as "heavy-use", those will be good candidates for an index(s). On Fri, Oct 24, 2014 at 7:58 AM, xuxc1120 <xuxc1...@vip.qq.com> wrote: Hi, i am sorry to conduct such question:i already have a hbase table with 1.5 million rows ,(e.g table1 with rowkey and column Family :info),under CF "info",these are almost 100 qualifies, such as "info:name","info:age","info:addr"....when I want to use index in phoenix to query data on any column,e.gselect * from table where "name"="cond1" and "addr"="cond2";select * from table where "age" ="cond3" and "name"= "cond4";...I want to use index to query for every possibility may select clause conduct .Are there some ways to do that?that is to say ,how to create table/view and indexes in phoenix ?thanks.xuxc
