Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Hive >> mail # user >> Thrift Server

Copy link to this message
Re: Thrift Server
Hi Ransom,
JDBC talks to the Hive driver via the Hive Thrift server. Since the thrift server is limited to single connection at a time, out of curiosity, how much benefit do concurrent connections through JDBC give you, given that they will also be handled serially when being sent to the Thrift server?

Thanks in advance!

----- Original Message -----
From: "Hezhiqiang (Ransom)" <[EMAIL PROTECTED]>
Cc: "Zhaojun (Terry)" <[EMAIL PROTECTED]>
Sent: Friday, June 29, 2012 9:31:01 PM
Subject: RE: Thrift Server

Hi Van
In my test, JDBC supported about 25 connections in the same time without delay.
If more connections , it will line up and timeout exception will happen.

-----Original Message-----
From: VanHuy Pham [mailto:[EMAIL PROTECTED]]
Sent: Saturday, June 30, 2012 8:59 AM
Subject: Re: Thrift Server

Thanks for the response.
I see. Would JDBC then be a better option for concurrent connections?
I am not aware of the implementation of hive-JDBC so wonder if it
support multiple connections?
Any idea?

On 6/29/12, Hezhiqiang (Ransom) <[EMAIL PROTECTED]> wrote:
> Hi Van
>          Hive doesn’t supported multiple connections now.
>          it's impossible for HiveServer to support concurrent connections
> using the current Thrift API
> you can see:
> https://issues.apache.org/jira/browse/HIVE-2935
> https://cwiki.apache.org/confluence/download/attachments/27362054/HiveServer2HadoopSummit2012BoF.pdf?version=1&modificationDate=1339790767000
> From: VanHuy Pham [mailto:[EMAIL PROTECTED]]
> Sent: Saturday, June 30, 2012 4:33 AM
> Subject: Thrift Server
> Hi hive folks,
>     Does hive thrift server support multiple requests from clients at the
> same time?
>     It looks like the server serves the request sequentially, which means it
> processes each request one by one. Am I wrong here?
>     I make two clients, which make two requests (select data) to two
> different in hive; judging by the terminal screen of the hive server, it
> processes one request, finishes it, and process the other.
> Van