[
https://issues.apache.org/jira/browse/HBASE-13291?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-13291:
--------------------------
Attachment: nonBBposAndInineMvccVint.txt
This is the hack to bypass bb plus inlining of the vint of mvcc. No longer
shows in perf top. Here is what it looks like now:
{code}
Samples: 4M of event 'cycles', Event count (approx.): 422533226118
21.09% perf-13042.map [.]
Lorg/apache/hadoop/hbase/io/hfile/HFileReaderV2$ScannerV2;._next in
Lorg/apache/hadoop/hbase/io/hfile/HFileReaderV2$ScannerV2;.next
12.76% perf-13042.map [.]
Lorg/apache/hadoop/hbase/regionserver/ScanQueryMatcher;.match
11.61% perf-13042.map [.] <A0>4=^D<F4>^?
6.00% perf-13042.map [.]
Lorg/apache/hadoop/hbase/regionserver/StoreScanner;.optimize in
Lorg/apache/hadoop/hbase/regionserver/StoreScanner;.next
5.84% perf-13042.map [.]
Lorg/apache/hadoop/hbase/regionserver/KeyValueHeap;.next
4.75% perf-13042.map [.]
Ljava/util/concurrent/locks/AbstractQueuedSynchronizer;.release in
Lorg/apache/hadoop/hbase/regionserver/StoreScanner;.peek
3.87% perf-13042.map [.]
Ljava/util/PriorityQueue;.siftUpUsingComparator in
Lorg/apache/hadoop/hbase/regionserver/KeyValueHeap;.next
3.57% perf-13042.map [.]
Lorg/apache/hadoop/hbase/regionserver/KeyValueHeap;.peek in
Lorg/apache/hadoop/hbase/regionserver/KeyValueHeap;.next
3.09% perf-13042.map [.]
Lorg/apache/hadoop/hbase/util/Bytes$LexicographicalComparerHolder$UnsafeComparer;.compareTo
in Lorg/apache/hadoop/hbase/regionserver/ScanQueryMatcher;.match
2.00% perf-13042.map [.]
Lorg/apache/hadoop/hbase/regionserver/ScanDeleteTracker;.reset in
Lorg/apache/hadoop/hbase/regionserver/StoreScanner;.next
1.96% libjvm.so [.]
BlockOffsetArrayNonContigSpace::block_start_unsafe(void const*) const
1.38% perf-13042.map [.]
Lorg/apache/hadoop/hbase/regionserver/StoreFileScanner;.next in
Lorg/apache/hadoop/hbase/regionserver/KeyValueHeap;.next
1.03% libjvm.so [.]
ClearNoncleanCardWrapper::do_MemRegion(MemRegion)
0.95% perf-13042.map [.] Ljava/util/concurrent/ConcurrentHashMap;.get
in Lorg/apache/hadoop/hbase/io/hfile/LruBlockCache;.getBlock
0.93% perf-13042.map [.]
Ljava/util/concurrent/locks/LockSupport;.unpark in
Lorg/apache/hadoop/hbase/regionserver/StoreScanner;.next
0.91% perf-13042.map [.] Ljava/nio/HeapByteBuffer;.slice in
Lorg/apache/hadoop/hbase/io/hfile/HFileReaderV2$ScannerV2;.updateCurrBlock
0.77% libjvm.so [.] typeArrayKlass::allocate_common(int, bool,
Thread*)
0.72% libjvm.so [.] CompactibleFreeListSpace::block_size(HeapWord
const*) const
0.60% perf-13042.map [.]
Lorg/apache/hadoop/hbase/util/Bytes$LexicographicalComparerHolder$UnsafeComparer;.compareTo
in
Lorg/apache/hadoop/hbase/regionserver/HRegion$RegionScannerImpl;.nextInternal
0.55% perf-13042.map [.] <D0>)=^D<F4>^?
0.49% libjvm.so [.] SpinPause
0.45% perf-13042.map [.]
Lorg/apache/hadoop/hbase/util/Bytes$LexicographicalComparerHolder$UnsafeComparer;.compareTo
in
Lorg/apache/hadoop/hbase/regionserver/HRegion$RegionScannerImpl;.populateResult
{code}
At about 8.5k/s. Up but not much.
> Lift the scan ceiling
> ---------------------
>
> Key: HBASE-13291
> URL: https://issues.apache.org/jira/browse/HBASE-13291
> Project: HBase
> Issue Type: Improvement
> Components: Scanners
> Affects Versions: 1.0.0
> Reporter: stack
> Assignee: stack
> Attachments: 13291.inlining.txt, Screen Shot 2015-03-26 at 12.12.13
> PM.png, Screen Shot 2015-03-26 at 3.39.33 PM.png, hack_to_bypass_bb.txt,
> nonBBposAndInineMvccVint.txt, q (1).png, traces.7.svg, traces.filterall.svg,
> traces.nofilter.svg, traces.small2.svg, traces.smaller.svg
>
>
> Scanning medium sized rows with multiple concurrent scanners exhibits
> interesting 'ceiling' properties. A server runs at about 6.7k ops a second
> using 450% of possible 1600% of CPUs when 4 clients each with 10 threads
> doing scan 1000 rows. If I add '--filterAll' argument (do not return
> results), then we run at 1450% of possible 1600% possible but we do 8k ops a
> second.
> Let me attach flame graphs for two cases. Unfortunately, there is some
> frustrating dark art going on. Let me try figure it... Filing issue in
> meantime to keep score in.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
