* Benchmarks: New SqlBenchmark, add caching & vectorization to some others.
- Introduce a new SqlBenchmark geared towards benchmarking a wide
variety of SQL queries. Rename the old SqlBenchmark to
SqlVsNativeBenchmark.
- Add (optional) caching to SegmentGenerator to enable easier
benchmarking of larger segments.
- Add vectorization to FilteredAggregatorBenchmark and GroupByBenchmark.
* Query vectorization.
This patch includes vectorized timeseries and groupBy engines, as well
as some analogs of your favorite Druid classes:
- VectorCursor is like Cursor. (It comes from StorageAdapter.makeVectorCursor.)
- VectorColumnSelectorFactory is like ColumnSelectorFactory, and it has
methods to create analogs of the column selectors you know and love.
- VectorOffset and ReadableVectorOffset are like Offset and ReadableOffset.
- VectorAggregator is like BufferAggregator.
- VectorValueMatcher is like ValueMatcher.
There are some noticeable differences between vectorized and regular
execution:
- Unlike regular cursors, vector cursors do not understand time
granularity. They expect query engines to handle this on their own,
which a new VectorCursorGranularizer class helps with. This is to
avoid too much batch-splitting and to respect the fact that vector
selectors are somewhat more heavyweight than regular selectors.
- Unlike FilteredOffset, FilteredVectorOffset does not leverage indexes
for filters that might partially support them (like an OR of one
filter that supports indexing and another that doesn't). I'm not sure
that this behavior is desirable anyway (it is potentially too eager)
but, at any rate, it'd be better to harmonize it between the two
classes. Potentially they should both do some different thing that
is smarter than what either of them is doing right now.
- When vector cursors are created by QueryableIndexCursorSequenceBuilder,
they use a morphing binary-then-linear search to find their start and
end rows, rather than linear search.
Limitations in this patch are:
- Only timeseries and groupBy have vectorized engines.
- GroupBy doesn't handle multi-value dimensions yet.
- Vector cursors cannot handle virtual columns or descending order.
- Only some filters have vectorized matchers: "selector", "bound", "in",
"like", "regex", "search", "and", "or", and "not".
- Only some aggregators have vectorized implementations: "count",
"doubleSum", "floatSum", "longSum", "hyperUnique", and "filtered".
- Dimension specs other than "default" don't work yet (no extraction
functions or filtered dimension specs).
Currently, the testing strategy includes adding vectorization-enabled
tests to TimeseriesQueryRunnerTest, GroupByQueryRunnerTest,
GroupByTimeseriesQueryRunnerTest, CalciteQueryTest, and all of the
filtering tests that extend BaseFilterTest. In all of those classes,
there are some test cases that don't support vectorization. They are
marked by special function calls like "cannotVectorize" or "skipVectorize"
that tell the test harness to either expect an exception or to skip the
test case.
Testing should be expanded in the future -- a project in and of itself.
Related to #3011.
* WIP
* Adjustments for unused things.
* Adjust javadocs.
* DimensionDictionarySelector adjustments.
* Add "clone" to BatchIteratorAdapter.
* ValueMatcher javadocs.
* Fix benchmark.
* Fixups post-merge.
* Expect exception on testGroupByWithStringVirtualColumn for IncrementalIndex.
* BloomDimFilterSqlTest: Tag two non-vectorizable tests.
* Minor adjustments.
* Update surefire, bump up Xmx in Travis.
* Some more adjustments.
* Javadoc adjustments
* AggregatorAdapters adjustments.
* Additional comments.
* Remove switching search.
* Only missiles.
Follow-up to #7223 that fixes a doc bug (a result-level cache property
was misspelled), changes the recommended "small cluster" threshold from
20 to 5 servers, and clarifies behavior of the various caching options.
* Broker backpressure.
Adds a new property "druid.broker.http.maxQueuedBytes" and a new context
parameter "maxQueuedBytes". Both represent a maximum number of bytes queued
per query before exerting backpressure on the channel to the data server.
Fixes#4933.
* Fix query context doc.
Updating the description of useCache
Updating query-context doc based on Gian's comment
Updating query-context doc based on Gian's comment
Updating query-context doc based on Gian's comment
Updating query-context doc based on Gian's comment
* Make timeout behavior consistent to document
* Refactoring BlockingPool and add more methods to QueryContexts
* remove unused imports
* Addressed comments
* Address comments
* remove unused method
* Make default query timeout configurable
* Fix test failure
* Change timeout from period to millis
* Ignore chunkPeriod for groupBy v2, fix chunkPeriod for irregular periods.
Includes two fixes:
- groupBy v2 now ignores chunkPeriod, since it wouldn't have helped anyway (its mergeResults
returns a lazy sequence) and it generates incorrect results.
- Fix chunkPeriod handling for periods of irregular length, like "P1M" or "P1Y".
Also includes doc and test fixes:
- groupBy v1 was no longer being tested by GroupByQueryRunnerTest since #3953, now it
is once again.
- chunkPeriod documentation was misleading due to its checkered past. Updated it to
be more accurate.
* Remove unused import.
* Restore buffer size.
* SQL: Add context and contextual functions to planner.
Added support for context parameters specified as JDBC connection properties
or a JSON object for SQL-over-JSON-over-HTTP.
Also added features that depend on context functionality:
- Added CURRENT_DATE, CURRENT_TIME, CURRENT_TIMESTAMP functions.
- Added support for time zones other than UTC via a "timeZone" context.
- Pass down query context to Druid queries too.
Also some bug fixes:
- Fix DATE handling, it was largely done incorrectly before.
- Fix CAST(__time TO DATE) which should do a floor-to-day.
- Fix non-equality comparisons to FLOOR(__time TO X).
- Fix maxQueryCount property.
* Pass down context to nested queries too.
* make isSingleThreaded groupBy query processing overridable at query time
* refactor code in GroupByMergedQueryRunner to make processing of single threaded and parallel merging of runners consistent
Gian Merlino
Jonathan Wei
Pierre-Emile Ferron
David Lim
Vadim Ogievetsky
Jihoon Son
Atul Mohan
Daniel
kaijianding
Himanshu
asrayousuf
binlijin