hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bryan Duxbury (JIRA)" <j...@apache.org>
Subject [jira] Created: (HBASE-632) HTable.getMetadata is very inefficient
Date Wed, 21 May 2008 14:57:55 GMT
HTable.getMetadata is very inefficient
--------------------------------------

                 Key: HBASE-632
                 URL: https://issues.apache.org/jira/browse/HBASE-632
             Project: Hadoop HBase
          Issue Type: Improvement
          Components: client
            Reporter: Bryan Duxbury
            Priority: Minor
             Fix For: 0.2.0


The current implementation of HTable.getMetadata is very inefficient. In order to find the
HTableDescriptor, it does a full meta scan for all table names. Then, it selects the one that
matches the table we're looking for. That's a huge number of operations to satisfy a very
simple question. To make matters worse, it doesn't even cache the result it gets - subsequent
calls will go right back out to a meta scan.

At the very least, we should only look up the first meta row for a given take and do it as
a random read. We should consider caching the result, or even just grabbing it during table
creation (as we will check meta to make sure the table you're opening exists). Of course,
if we move table descriptors out of META altogether (maybe into Zookeeper?) then this will
have to be revisited no matter what.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message