[
https://issues.apache.org/jira/browse/HDFS-8668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15408477#comment-15408477
]
Kai Zheng commented on HDFS-8668:
---------------------------------
Thanks [~Sammi] for resuming this task. The updated patch looks good overall.
{code}
+ /**
+ * This was because we don't have appropriate efficient ByteBuffer
version
+ * downstream calls, so we have to convert to heap buffer to proceed.
+ * We can consider to optimize this separately into the underlying layer.
+ */
+ heapBuffer = BUFFER_POOL.getBuffer(false, buffer.remaining());
+ buffer.get(heapBuffer.array());
+ buffer.position(0);
{code}
The situation indicated by the comment is obsolete and not true, since
{{DataChecksum}} is of {{calculateChunkedSums(ByteBuffer data, ByteBuffer
checksums)}}. Could you help check if the buffer converting is really needed?
Better to avoid the buffer copy for performance. Thanks.
> Erasure Coding: revisit buffer used for encoding and decoding.
> --------------------------------------------------------------
>
> Key: HDFS-8668
> URL: https://issues.apache.org/jira/browse/HDFS-8668
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Yi Liu
> Assignee: SammiChen
> Attachments: HDFS-8668-v1.patch, HDFS-8668-v2.patch,
> HDFS-8668-v3.patch
>
>
> For encoding and decoding buffers, currently some places use java heap
> ByteBuffer, some use direct byteBUffer, and some use java byte array. If
> the coder implementation is native, we should use direct ByteBuffer. This
> jira is to revisit all encoding/decoding buffers and improve them.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
