MySQL 的可扩展性如何针对 BenchmarkSQL TPC-C 测试进行优化？

commit 6be2fa0bdbbadc52cc8478b52b69db02b0eaff40
Author: Paweł Olchawa <[email protected]>
Date:   Wed Feb 14 09:33:42 2018 +0100
    
    WL#10310 Redo log optimization: dedicated threads and concurrent log buffer.

    0. Log buffer became a ring buffer, data inside is no longer shifted.
    1. User threads are able to write concurrently to log buffer. 
    2. Relaxed order of dirty pages in flush lists - no need to synchronize
       the order in which dirty pages are added to flush lists.
    3. Concurrent MTR commits can interleave on different stages of commits.
    4. Introduced dedicated log threads which keep writing log buffer: 
        * log_writer: writes log buffer to system buffers,
        * log_flusher: flushes system buffers to disk.
       As soon as they finished writing (flushing) and there is new data to 
       write (flush), they start next write (flush).
    5. User threads no longer write / flush log buffer to disk, they only
       wait by spinning or on event for notification. They do not have to 
       compete for the responsibility of writing / flushing.
    6. Introduced a ring buffer of events (one per log-block) which are used
       by user threads to wait for written/flushed redo log to avoid:
        * contention on single event
        * false wake-ups of all waiting threads whenever some write/flush
          has finished (we can wake-up only those waiting in related blocks)
    7. Introduced dedicated notifier threads not to delay next writes/fsyncs:
        * log_write_notifier: notifies user threads about written redo,
        * log_flush_notifier: notifies user threads about flushed redo.
    8. Master thread no longer has to flush log buffer.
    ...
    30. Mysql test runner received a new feature (thanks to Marcin):
        --exec_in_background.
    Review: RB#15134
    Reviewers:
        - Marcin Babij <[email protected]>,
        - Debarun Banerjee <[email protected]>.
    Performance tests:
        - Dimitri Kravtchuk <[email protected]>,
        - Daniel Blanchard <[email protected]>,
        - Amrendra Kumar <[email protected]>.
    QA and MTR tests:
        - Vinay Fisrekar <[email protected]>.

commit bc95476c0156070fd5cedcfd354fa68ce3c95bdb
Author: Paweł Olchawa <[email protected]>
Date:   Tue May 25 18:12:20 2021 +0200
    
    BUG#32832196 SINGLE RW_TRX_SET LEADS TO CONTENTION ON TRX_SYS MUTEX
    
    1. Introduced shards, each with rw_trx_set and dedicated mutex.
    2. Extracted modifications to rw_trx_set outside its original critical sections
       (removal had to be extracted outside trx_erase_lists).
    3. Eliminated allocation on heap inside TrxUndoRsegs.
    4. [BUG-FIX] The trx->state and trx->start_time became converted to std::atomic<>
       fields to avoid risk of torn reads on egzotic platforms.
    5. Added assertions which ensure that thread operating on transaction has rights
       to do so (to show there is no possible race condition).
    
    RB: 26314
    Reviewed-by: Jakub Łopuszański [email protected]