Mailing-List: contact dev-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hbase.apache.org
Date: Fri, 29 Jul 2016 17:17:20 +0000 (UTC)
From: "Elliott Clark (JIRA)" <jira@apache.org>
To: dev@hbase.apache.org
Message-ID: <JIRA.12993596.1469812581000.170760.1469812640515@Atlassian.JIRA>
In-Reply-To: <JIRA.12993596.1469812581000@Atlassian.JIRA>
References: <JIRA.12993596.1469812581000@Atlassian.JIRA> <JIRA.12993596.1469812581562@arcas>
Subject: [jira] [Created] (HBASE-16305) Multi threaded scan performance
 bottlenecked on Connection
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Fri, 29 Jul 2016 17:17:23 -0000

Elliott Clark created HBASE-16305:
-------------------------------------

             Summary: Multi threaded scan performance bottlenecked on Connection
                 Key: HBASE-16305
                 URL: https://issues.apache.org/jira/browse/HBASE-16305
             Project: HBase
          Issue Type: Bug
            Reporter: Elliott Clark


Pretty simple repro. On a single region server host ~10 or more regions. Then for each region spawn a thread that starts a scan. Try an pull the scan as fast as possible.

If all the scans are created through the same connection then it's slow. If all the scans are created through different connections then the scan speed is doubled.

Since we recommend a single connection per application this is pretty surprising behavior.


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)