[
https://issues.apache.org/jira/browse/HBASE-82?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12569252#action_12569252
]
Jim Kellerman commented on HBASE-82:
------------------------------------
I've been doing some thinking about this and would like to address some
arguments that have been made both for and against this change:
With respect to custom WritableComparables and Comparators;
- It would not be too hard to get custom code to the servers. It just means
that the custom code has to be distributed to every node running HBase and the
classpath has to be adjusted. This is also a requirement for using custom
classes in Map/Reduce, so people are already familiar with doing this.
- However, having said that, I do agree that serialization and deserialization
of custom classes is much more expensive than for a restricted set of "known"
classes.
With respect to using byte arrays as keys:
- This is appealing to me. Serialization and deserialization are easy and fast
as is comparison. It would be easy to extend ImmutableBytesWritable to
implement Comparable.
- BerkeleyDB uses byte arrays as keys and values and allows either to be up to
4GB (unsigned int) in length. We could live with the 2GB limit that Java's
signed int imposes.
The biggest downside to changing keys from Text to anything else is migration.
However:
- We do have a migration framework in place.
- If we are going to make this change it would be better to do it sooner rather
than later before users have hundreds of GB of data stored in HBase.
- It is easy to convert between Text and byte
{code}
bytes[] b = something
Text t = new Text(b)
bytes[] b = t.getBytes();
t.set(b);
{code}
> [hbase] VOTE: should row keys be less restrictive than hadoop.io.Text?
> ----------------------------------------------------------------------
>
> Key: HBASE-82
> URL: https://issues.apache.org/jira/browse/HBASE-82
> Project: Hadoop HBase
> Issue Type: Wish
> Reporter: Jim Kellerman
> Priority: Minor
>
> I have heard from several people that row keys in HBase should be less
> restricted than hadoop.io.Text.
> What do you think?
> At the very least, a row key has to be a WritableComparable. This would lead
> to the most general case being either hadoop.io.BytesWritable or
> hbase.io.ImmutableBytesWritable. The primary difference between these two
> classes is that hadoop.io.BytesWritable by default allocates 100 bytes and if
> you do not pay attention to the length, (BytesWritable.getSize()), converting
> a String to a BytesWritable and vice versa can become problematic.
> hbase.io.ImmutableBytesWritable, in contrast only allocates as many bytes as
> you pass in and then does not allow the size to be changed.
> If we were to change from Text to a non-text key, my preference would be for
> ImmutableBytesWritable, because it has a fixed size once set, and operations
> like get, etc do not have to something like System.arrayCopy where you
> specify the number of bytes to copy.
> Your comments, questions are welcome on this issue. If we receive enough
> feedback that Text is too restrictive, we are willing to change it, but we
> need to hear what would be the most useful thing to change it to as well.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
