KunpengBoostKit22.0.RC4.ThreadCounterOptimization for MySQL 8.0.25

Optimize performance of status variable thread_running in high concurrency scenarios like sysbench point select.

KunpengBoostKit22.0.RC4 threadpool for MySQL8.0.25

Improved performance of the thread pool plugin when high concurrency.

KunpengBoostKit22.0.RC4.SchedAffinity

This patch offers better control of MySQL 8.0.25 thread scheduling on multiprocessor architecture. Specifically,

1. To take advantages of data locality, foreground threads can be set to run in a target NUMA node.
2. To avoid key background threads become the bottleneck, these threads can be set to run in their dedicated cores.

KunpengBoostKit22.0.RC2 threadpool for MySQL8.0.25

Support thread pool connection handler in MySQL8.0.25.we re-implemented the thread pool as a plugin.

KunpengBoostKit22.0.RC2 parallel query patch for MySQL8.0.25

parallel query patch for MySQL8.025,support parallel query on single table or multiple tables, do not support subquery.
support tpch query:Q1 Q3 Q4 Q5 Q6 Q7 Q9 Q10 Q12 Q19 Q21

KunpengBoostKit21.0.0 threadpool for MySQL5

Support thread pool connection handler in MySQL5.7.27.

KunpengBoostKit21.0.0 threadpool for MySQL8

Support thread pool connection handler in MySQL8.0.20.

KunpengBoostKit21.0.0 OLTP

Optimize performance of status variable thread_running in high concurrency scenarios like sysbench point select.

KunpengBoostKit21.0.RC1 Parallel Query Patch

parallel query patch,support parallel query on single table or multiple tables, do not support subquery.
support tpch query:Q1 Q3 Q4 Q5 Q6 Q7 Q9 Q10 Q12 Q19 Q21

KunpengBoostKit21.0.RC1

• The 0001-HASH-TABLE-LOCKS-OPT.patch reduce contention on hash_table_locks.
• The 0001-UNDO-SPACES-LOCK-OPT.patch reduce contention on undo_spaces_lock.
• The 0002-UNDO-SPACES-LOCK-OPT.AFTER-SCHED-AFFINITY.patch is the same as 0001-UNDO-SPACES-LOCK-OPT.patch execpt it has resolved confliction with 0001-SCHED-AFFINITY.patch.
• The 0001-THREAD_POOL.patch provides thread pool feature, that enables MySQL to serve a large number of client connections while keeping a good perfomance.