[ https://issues.apache.org/jira/browse/HIVE-3992?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13573402#comment-13573402 ]
Gopal V commented on HIVE-3992: ------------------------------- Testing dummy query (to simulate a "col in (select ...)" style query) at SCALE=10 select /*+MAPJOIN(time_dim)*/ store_sales_rc.ss_item_sk from store_sales_rc join time_dim on (store_sales_rc.ss_sold_time_sk = time_dim.t_time_sk) limit 100; Before {code} 2013-02-07 06:32:02,164 Stage-1 map = 0%, reduce = 0% 2013-02-07 06:32:20,082 Stage-1 map = 50%, reduce = 0%, Cumulative CPU 53.9 sec 2013-02-07 06:32:21,127 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 61.59 sec Job 0: Map: 8 Cumulative CPU: 61.59 sec HDFS Read: 104763092 HDFS Write: 4749 SUCCESS Total MapReduce CPU Time Spent: 1 minutes 1 seconds 590 msec Time taken: 34.572 seconds, Fetched: 100 row(s) {code} After {code} 2013-02-07 06:35:29,413 Stage-1 map = 0%, reduce = 0% 2013-02-07 06:35:43,200 Stage-1 map = 25%, reduce = 0%, Cumulative CPU 9.31 sec 2013-02-07 06:35:44,247 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 39.45 sec MapReduce Total cumulative CPU time: 39 seconds 450 msec Ended Job = job_1359695160319_0164 MapReduce Jobs Launched: Job 0: Map: 8 Cumulative CPU: 39.45 sec HDFS Read: 25416952 HDFS Write: 4749 SUCCESS Total MapReduce CPU Time Spent: 39 seconds 450 msec Time taken: 31.351 seconds, Fetched: 100 row(s) {code} Now the interesting bit is that even though we cut down the CPU cost by almost 50%, the over-all latency drops only by 2 secs. > Hive RCFile::sync(long) does a sub-sequence linear search for sync blocks > ------------------------------------------------------------------------- > > Key: HIVE-3992 > URL: https://issues.apache.org/jira/browse/HIVE-3992 > Project: Hive > Issue Type: Bug > Environment: Ubuntu x86_64/java-1.6/hadoop-2.0.3 > Reporter: Gopal V > Attachments: HIVE-3992.patch, select-join-limit.html > > > The following function does some bad I/O > {code} > public synchronized void sync(long position) throws IOException { > ... > try { > seek(position + 4); // skip escape > in.readFully(syncCheck); > int syncLen = sync.length; > for (int i = 0; in.getPos() < end; i++) { > int j = 0; > for (; j < syncLen; j++) { > if (sync[j] != syncCheck[(i + j) % syncLen]) { > break; > } > } > if (j == syncLen) { > in.seek(in.getPos() - SYNC_SIZE); // position before > // sync > return; > } > syncCheck[i % syncLen] = in.readByte(); > } > } > ... > } > {code} > This causes a rather large number of readByte() calls which are passed onto a > ByteBuffer via a single byte array. > This results in rather a large amount of CPU being burnt in a the linear > search for the sync pattern in the input RCFile (upto 92% for a skewed > example - a trivial map-join + limit 100). > This behaviour should be avoided at best or at least replaced by a rolling > hash for efficient comparison, since it has a known byte-width of 16 bytes. > Attached the stack trace from a Yourkit profile. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira