答复: Regarding query compile

[Wangwenli <wa...@huawei.com>]

Thanks yours quick response, i raised one issue:PHOENIX-961,  and provide some infomation, pls let me know whether need more info. also, i will check more for this.  Thanks Again


Regards
Wenli

________________________________
发件人: James Taylor [jamestaylor@apache.org]
发送时间: 2014年4月30日 1:37
收件人: user@phoenix.incubator.apache.org
主题: Re: Regarding query compile

Hi Wenli,
We've typically not seen the compilation of the query as being any significant overhead when compared with all the HBase-related activity when a query is run (serialization, RPC, deserialization, scanning, etc), hence we have not attempted to re-use the QueryPlan. It would be possible, though. Can you share with us some performance numbers that breaks down the time spent in each phase (mainly query compilation versus query execution)? Please open a JIRA for this.
Thanks,
James


On Tue, Apr 29, 2014 at 6:30 AM, Wangwenli <wa...@huawei.com>> wrote:

HI James and others ,

    currently, for each statement.executequery, even the query is same ,just Parameters is different, the statement will be compile everytime,  it seems other than the dynamic parameters, other part like ColumnResolver,GroupBy etc all of these are same, why cann't these be reused?  i mean the ScanPlan can be reused between the same statement.
   it seems, currently we found these much of complie cost too much cpu time.

   got my question?


Regards
Wenli