[
https://issues.apache.org/jira/browse/TRAFODION-2917?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16360294#comment-16360294
]
ASF GitHub Bot commented on TRAFODION-2917:
-------------------------------------------
Github user sureshsubbiah commented on a diff in the pull request:
https://github.com/apache/trafodion/pull/1417#discussion_r167412178
--- Diff: core/sql/executor/ExHdfsScan.cpp ---
@@ -118,15 +119,39 @@ ExHdfsScanTcb::ExHdfsScanTcb(
, dataModCheckDone_(FALSE)
, loggingErrorDiags_(NULL)
, loggingFileName_(NULL)
+ , hdfsClient_(NULL)
+ , hdfsScan_(NULL)
+ , hdfsStats_(NULL)
, hdfsFileInfoListAsArray_(glob->getDefaultHeap(),
hdfsScanTdb.getHdfsFileInfoList()->numEntries())
{
Space * space = (glob ? glob->getSpace() : 0);
CollHeap * heap = (glob ? glob->getDefaultHeap() : 0);
+ useLibhdfsScan_ = hdfsScanTdb.getUseLibhdfsScan();
+ if (isSequenceFile())
+ useLibhdfsScan_ = TRUE;
lobGlob_ = NULL;
- const int readBufSize = (Int32)hdfsScanTdb.hdfsBufSize_;
- hdfsScanBuffer_ = new(space) char[ readBufSize + 1 ];
- hdfsScanBuffer_[readBufSize] = '\0';
-
+ hdfsScanBufMaxSize_ = hdfsScanTdb.hdfsBufSize_;
+ headRoom_ = (Int32)hdfsScanTdb.rangeTailIOSize_;
+
+ if (useLibhdfsScan_) {
+ hdfsScanBuffer_ = new(heap) char[ hdfsScanBufMaxSize_ + 1 ];
+ hdfsScanBuffer_[hdfsScanBufMaxSize_] = '\0';
+ } else {
+ hdfsScanBufBacking_[0] = new (heap) BYTE[hdfsScanBufMaxSize_ + 2 *
(headRoom_)];
--- End diff --
Could we please test this logic with extremely wide rows. Currently we have
a limitation that the maximum row size cannot be larger than
hdfsScanBufMaxSize_. It will be a good test to have 10 rows of this size, say
in 2 or more files and check if we can process it. Logic seems good, this is a
test suggestion. As you know scanBufNaxSize can re reduced with a cqd, to avoid
having to deal with rows that are several MBs wide.
> Refactor Trafodion implementation of hdfs scan for text formatted hive tables
> -----------------------------------------------------------------------------
>
> Key: TRAFODION-2917
> URL: https://issues.apache.org/jira/browse/TRAFODION-2917
> Project: Apache Trafodion
> Issue Type: New Feature
> Components: sql-general
> Reporter: Selvaganesan Govindarajan
> Priority: Major
> Fix For: 2.3
>
>
> Find below the general outline of hdfs scan for text formatted hive tables.
> Compiler returns a list of scan ranges and the begin range and number of
> ranges to be done by each instance of TCB in TDB. This list of scan ranges is
> also re-computed at run time possibly based on a CQD
> The scan range for a TCB can come from the same or different hdfs files. TCB
> creates two threads to read these ranges.Two ranges (for the TCB) are
> initially assigned to these threads. As and when a range is completed, the
> next range (assigned for the TCB) is picked up by the thread. Ranges are read
> in multiples of hdfs scan buffer size at the TCB level. Default hdfs scan
> buffer size is 64 MB. Rows from hdfs scan buffer is processed and moved into
> up queue. If the range contains a record split, then the range is extended to
> read up to range tail IO size to get the full row. The range that had the
> latter part of the row ignores it because the former range processes it.
> Record split at the file level is not possible and/or not supported.
> For compression, the compiler returns the range info such that the hdfs scan
> buffer can hold the full uncompressed buffer.
> Cons:
> Reader threads feature too complex to maintain in C++
> Error handling at the layer below the TCB is missing or errors are not
> propagated to work method causing incorrect results
> Possible multiple copying of data
> Libhdfs calls are not optimized. It was observed that the method Ids are
> being obtained many times. Need to check if this problem still exists.
> Now that we clearly know what is expected, it could be optimized better
> - Reduced scan buffer size for smoother data flow
> - Better thread utilization
> - Avoid multiple copying of data.
> Unable to comprehend the need for two threads for pre-fetch especially when
> one range is completed fully before the data from next range is processed.
> Following are the hdfsCalls used by programs at exp and executor directory.
> U hdfsCloseFile
> U hdfsConnect
> U hdfsDelete
> U hdfsExists
> U hdfsFlush
> U hdfsFreeFileInfo
> U hdfsGetPathInfo
> U hdfsListDirectory
> U hdfsOpenFile
> U hdfsPread
> U hdfsRename
> U hdfsWrite
> U hdfsCreateDirectory
> New implementation
> Make changes to use direct Java APIs for these calls. However, come up with
> better mechanism to move the data from Java and JNI, avoid unnecessary
> copying of data, better thread management via Executor concepts in Java.
> Hence it won’t be direct mapping of these calls to hdfs Java API. Instead,
> use the abstraction like what is being done for HBase access.
> I believe newer implementation will be optimized better and hence improved
> performance. (but not many folds)
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
