ClickHouse® is a real-time analytics DBMS
Go to file
Azat Khuzhin 35231662b3 Improve performance of AggregatingMergeTree w/ SimpleAggregateFunction(String)
While reading from AggregatingMergeTree with
SimpleAggregateFunction(String) in primary key and
optimize_aggregation_in_order perf top shows:

    Samples: 1M of event 'cycles', 4000 Hz, Event count (approx.): 287759760270 lost: 0/0 drop: 0/0
      Children      Self  Shared Object         Symbol
    +   12.64%    11.39%  clickhouse            [.] memcpy
    +    9.08%     0.23%  [unknown]             [.] 0000000000000000
    +    8.45%     8.40%  clickhouse            [.] ProfileEvents::increment    # <-- this, and in debug it has not 0.08x overhead, but 5.8x overhead
    +    7.68%     7.67%  clickhouse            [.] LZ4_compress_fast_extState
    +    5.29%     5.22%  clickhouse            [.] DB::IAggregateFunctionHelper<DB::AggregateFunctionNullUnary<true, true> >::addFree

The reason is obvious, ProfileEvents is atomic counters (and also they
are nested):

<details>

```
    Samples: 7M of event 'cycles', 4000 Hz, Event count (approx.): 450726149337
    ProfileEvents::increment  /usr/bin/clickhouse [Percent: local period]
    Percent│
           │
           │
           │    Disassembly of section .text:
           │
           │    00000000078d8900 <ProfileEvents::increment(unsigned long, unsigned long)@@Base>:
           │    ProfileEvents::increment(unsigned long, unsigned long):
      0.17 │      push  %rbp
      0.00 │      mov   %rsi,%rbp
      0.04 │      push  %rbx
      0.20 │      mov   %rdi,%rbx
      0.17 │      sub   $0x8,%rsp
      0.26 │    → callq DB::CurrentThread::getProfileEvents
           │    ProfileEvents::Counters::increment(unsigned long, unsigned long):
      0.00 │      lea   0x0(,%rbx,8),%rdi
      0.05 │      nop
           │    unsigned long std::__1::__cxx_atomic_fetch_add<unsigned long, unsigned long>(std::__1::__cxx_atomic_base_impl<unsigned long>*, unsigned long, std::__1::memory_order):
      1.02 │      mov   (%rax),%rdx
     97.04 │      lock  add   %rbp,(%rdx,%rdi,1)
           │    ProfileEvents::Counters::increment(unsigned long, unsigned long):
      0.21 │      mov   0x10(%rax),%rax
      0.04 │      test  %rax,%rax
      0.00 │    → jne   78d8920 <ProfileEvents::increment(unsigned long, unsigned long)@@Base+0x20>
           │    ProfileEvents::increment(unsigned long, unsigned long):
      0.38 │      add   $0x8,%rsp
      0.00 │      pop   %rbx
      0.04 │      pop   %rbp
      0.38 │    ← retq
```

</details>

These ProfileEvents was ArenaAllocChunks (it shows ~1.5M events per
second), and the reason is that the table has
SimpleAggregateFunction(String) in PK, which requires Arena.
But most of the time there Arena wasn't even used, so avoid this cost by
re-creating Arena only if it was "used" (i.e. has new chunks).

Another possibility is to avoid populating Arena::head in ctor, but this
will make the Arena code more complex, so for now this was preferred.

Also as a long-term solution it worth looking at implementing them via
RCU (to move the extra overhead out from the write code path into read
side).
2020-11-19 23:06:12 +03:00
.github Add experimental codecov.yml 2020-09-30 12:51:07 +03:00
base Merge pull request #17108 from ClickHouse/ya-make-owners 2020-11-17 05:19:46 +03:00
benchmark Convert to python3 (#15007) 2020-10-02 19:54:07 +03:00
cmake improve unbundle build 2020-11-11 09:08:53 +08:00
contrib Fix the case when CFA register is RAX 2020-11-15 18:32:41 +03:00
debian Auto version update to [20.12.1.1] [54443] 2020-11-05 21:52:50 +03:00
docker Moving to using TestFlows version 1.6.65 2020-11-17 20:22:43 -05:00
docs DOCSUP-3845: documented new encrypt and decrypt functions (#16930) 2020-11-17 17:05:33 +03:00
programs Follow Arcadia ya.make rules 2020-11-16 21:24:58 +03:00
src Improve performance of AggregatingMergeTree w/ SimpleAggregateFunction(String) 2020-11-19 23:06:12 +03:00
tests Improve performance of AggregatingMergeTree w/ SimpleAggregateFunction(String) 2020-11-19 23:06:12 +03:00
utils Follow Arcadia ya.make rules 2020-11-16 21:24:58 +03:00
website Added results from Raspberry Pi 2020-11-14 23:37:32 +03:00
.arcignore
.clang-format
.clang-tidy
.editorconfig
.gitattributes Better stats on github [#METR-20000]. 2016-05-04 18:03:17 +03:00
.gitignore Fix build without libraries 2020-10-10 23:41:27 +03:00
.gitmodules Merge branch 'master' into storage-rocksdb 2020-11-12 08:26:27 +08:00
.potato.yml
.vimrc
AUTHORS
CHANGELOG.md 20.8 2020-11-13 20:43:05 +03:00
CMakeLists.txt Add StorageEmbeddedRocksdb Engine 2020-11-11 09:08:52 +08:00
CODE_OF_CONDUCT.md
CONTRIBUTING.md
docker-compose.yml
format_sources
LICENSE
PreLoad.cmake
README.md Update README.md 2020-10-27 20:40:54 +03:00
release Convert to python3 (#15007) 2020-10-02 19:54:07 +03:00
SECURITY.md Update SECURITY.md 2020-10-01 19:17:11 +03:00
uncrustify.cfg Better .clang-format and uncrustify.cfg 2018-11-29 15:45:34 +03:00
ya.make

ClickHouse — open source distributed column-oriented DBMS

ClickHouse is an open-source column-oriented database management system that allows generating analytical data reports in real time.

  • Official website has quick high-level overview of ClickHouse on main page.
  • Tutorial shows how to set up and query small ClickHouse cluster.
  • Documentation provides more in-depth information.
  • YouTube channel has a lot of content about ClickHouse in video format.
  • Slack and Telegram allow to chat with ClickHouse users in real-time.
  • Blog contains various ClickHouse-related articles, as well as announcements and reports about events.
  • Code Browser with syntax highlight and navigation.
  • Yandex.Messenger channel shares announcements and useful links in Russian.
  • Contacts can help to get your questions answered if there are any.
  • You can also fill this form to meet Yandex ClickHouse team in person.

Upcoming Events