Refine the documentation and references further according to the recommendations.

wangbin579 · wangbin579 · commit 284f96aa3ed1 · 2025-04-23T09:30:03.000+08:00
diff --git a/Appendix.md b/Appendix.md
@@ -274,7 +274,7 @@ This schema is used by five different transactions, each creating varied access
 4. **Order and Order-Line:** Inserts with time-delayed updates, causing rows to become stale and infrequently read.
 5. **History:** Insert-only.
 
-The diverse access patterns of this small schema with a limited number of transactions contribute to TPC-C's ongoing significance as a major database benchmark. In this book, BenchmarkSQL [65] is primarily employed to evaluate TPC-C performance in MySQL.
+The diverse access patterns of this small schema with a limited number of transactions contribute to TPC-C's ongoing significance as a major database benchmark. In this book, BenchmarkSQL [68] is primarily employed to evaluate TPC-C performance in MySQL.
 
 ## How MySQL Processes SQL?
 
@@ -652,6 +652,6 @@ For MySQL clusters, the patch introduces further optimizations for **Group Repli
 
 ## About the Author
 
-In earlier years, Bin Wang worked at an internet company focused on developing high-performance computing and high-concurrency systems. He also contributed to open-source projects like TCPCopy [60] and MySQL Proxy [61], gaining valuable experience in problem-solving, particularly in logical thinking.
+In earlier years, Bin Wang worked at an internet company focused on developing high-performance computing and high-concurrency systems. He also contributed to open-source projects like TCPCopy [65] and MySQL Proxy [66], gaining valuable experience in problem-solving, particularly in logical thinking.
 
-After leaving the internet company, he concentrated on MySQL-related development, successfully contributing to projects such as Group Replication, secondary replay, InnoDB storage engines, and query optimization [64]. He has accumulated extensive experience in problem-solving within the MySQL domain.
+After leaving the internet company, he concentrated on MySQL-related development, successfully contributing to projects such as Group Replication, secondary replay, InnoDB storage engines, and query optimization [67]. He has accumulated extensive experience in problem-solving within the MySQL domain.
diff --git a/Chapter4_1.md b/Chapter4_1.md
@@ -36,7 +36,7 @@ The figure below illustrates the comparison results of TPC-C tests across differ
 
 Figure 4-3. Performance Comparison in SMP vs. NUMA.
 
-In the scenario where NUMA node 0 is bound, the throughput versus concurrency curve is notably smooth. Even under high concurrency, there is only a slight decline in throughput, indicating low thread context switching costs. However, throughput consistently remains below 400,000 tpmC due to significant limitations in memory bandwidth, characteristic of traditional SMP architecture.
+In the scenario where NUMA node 0 is bound, the throughput versus concurrency curve is notably smooth. Even under high concurrency, there is only a slight decline in throughput, indicating low thread context switching costs [60]. However, throughput consistently remains below 400,000 tpmC due to significant limitations in memory bandwidth, characteristic of traditional SMP architecture.
 
 In contrast, when utilizing all NUMA nodes, the throughput curve is relatively worse. This is attributed to reduced memory efficiency and increased context switching costs when accessing across NUMA nodes, resulting in less stable throughput. Nevertheless, scalability is greatly improved, with peak throughput increasing by 123% compared to using a single NUMA node.
 
@@ -48,8 +48,8 @@ Many pieces of code are not suitable for NUMA environments. For example, frequen
 
 To achieve optimal performance on NUMA systems [4], the following strategies are crucial:
 
-1.  Maximize the proportion of memory accesses routed to local nodes.
-2.  Balance traffic across nodes and interconnect links.
+1. Maximize the proportion of memory accesses routed to local nodes.
+2. Balance traffic across nodes and interconnect links.
 
 An unbalanced distribution of memory requests can significantly increase memory access latency on overloaded controllers, sometimes reaching up to 1000 cycles compared to approximately 200 cycles on non-overloaded controllers.
 
diff --git a/Chapter4_12.md b/Chapter4_12.md
@@ -26,7 +26,7 @@ The most efficient testing method is to utilize real online traffic for evaluati
 
 In Oracle, Database Replay enables testing a system with real production workloads, helping identify potential problems before implementing changes on the production system. Any workload period can be captured with little overhead and used to drive a test system, maintaining the concurrency and load characteristics of the real workload. Maintaining these characteristics is crucial, as current testing solutions often lack synchronization based on data dependencies. Without proper synchronization, the workload does not perform as required, leading to poor coverage and inadequate load, leaving many problems undetected. Database Replay's data-based synchronization makes testing realistic and helps discover potential problems [34].
 
-In MySQL, a common strategy involves taking a MySQL secondary instance offline for testing, configuring the necessary cluster, and replicating online MySQL requests to this new testing primary. The closer the testing primary resembles the production environment, the more accurate the test results. There are various methods to replicate online MySQL requests. This book recommends the open-source tool TCPCopy [60]. By using TCPCopy, many online problems have been successfully effectively resolved, laying a solid foundation for MySQL proxy enhancements [61]. For testing a MySQL cluster, replicating online requests to the testing system using TCPCopy allows us to evaluate whether the modifications achieve the expected outcomes, such as performance improvements, and robustness.
+In MySQL, a common strategy involves taking a MySQL secondary instance offline for testing, configuring the necessary cluster, and replicating online MySQL requests to this new testing primary. The closer the testing primary resembles the production environment, the more accurate the test results. There are various methods to replicate online MySQL requests. This book recommends the open-source tool TCPCopy [65]. By using TCPCopy, many online problems have been successfully effectively resolved, laying a solid foundation for MySQL proxy enhancements [66]. For testing a MySQL cluster, replicating online requests to the testing system using TCPCopy allows us to evaluate whether the modifications achieve the expected outcomes, such as performance improvements, and robustness.
 
 ### 4.12.5 Is Testing About Discovering Problems or Verifying Them?
 
diff --git a/Chapter4_2.md b/Chapter4_2.md
@@ -196,6 +196,8 @@ class hash_table_t {
   ...
 ```
 
+Note that the page-based buffer pool has low caching efficiency, and the page translation table is a scalability bottleneck [64]. 
+
 The certification database of Group Replication uses the *std::unordered_map* hash table to handle a large volume of certification information.
 
 ```c++
diff --git a/Chapter8.md b/Chapter8.md
@@ -145,7 +145,7 @@ From the figure, it is evident that after applying the patch, the rate of throug
 
 Addressing this problem directly presents considerable challenges, particularly for MySQL developers unfamiliar with query execution plans. Using logical reasoning and a systematic approach to identify and address code differences before and after the problem arose is a more elegant problem-solving method, though it is complex.
 
-It is noteworthy that no regression testing problems were encountered after applying the patch, demonstrating high stability and providing a solid foundation for future performance improvements. Currently, MySQL 8.0.40 still hasn't solved this problem, suggesting potential shortcomings in MySQL's testing system. Given the complexity of MySQL databases, users should exercise caution when upgrading and consider using tools like TCPCopy [60] to avoid potential regression testing problems.
+It is noteworthy that no regression testing problems were encountered after applying the patch, demonstrating high stability and providing a solid foundation for future performance improvements. Currently, MySQL 8.0.40 still hasn't solved this problem, suggesting potential shortcomings in MySQL's testing system. Given the complexity of MySQL databases, users should exercise caution when upgrading and consider using tools like TCPCopy [65] to avoid potential regression testing problems.
 
 ### 8.1.2 Improving Binlog Group Commit Scalability
 
@@ -633,7 +633,7 @@ Based on extensive testing, after solving most of MySQL's scalability problems,
 
 Centralized databases struggle to fully utilize hundreds of CPU cores due to limitations in their transaction systems. To address this, transaction throttling mechanisms are becoming increasingly important.
 
-MySQL has introduced a "Max Transaction Limit" feature in its thread pool to mitigate performance degradation [31]. This feature limits the number of concurrent transactions, improving throughput by reducing data locks and deadlocks on heavily loaded systems. This approach can inspire similar mechanisms that increase throughput in high-concurrency scenarios without relying solely on traditional thread pools.
+MySQL has introduced a "Max Transaction Limit" feature in its thread pool to mitigate performance degradation [31]. This feature limits the number of concurrent transactions, improving throughput by reducing data locks and deadlocks on heavily loaded systems [64]. This approach can inspire similar mechanisms that increase throughput in high-concurrency scenarios without relying solely on traditional thread pools.
 
 For MySQL, the specific process figure for transaction throttling is as follows:
 
diff --git a/Preface.md b/Preface.md
@@ -49,7 +49,7 @@ Part 5 is the concluding summary. Chapter 12 outlines future directions for MySQ
 
 ## References and Further Reading
 
-This book focuses on analyzing and solving MySQL problems, so a certain level of computer science background is recommended. To support understanding and maintain continuity, key terminology is included in the "Glossary" section of the appendix. For those lacking a foundation in MySQL, please refer to the related content in the appendix or consult dedicated MySQL books.
+This book focuses on analyzing and solving MySQL problems, so a certain level of computer science background is recommended. To support understanding and maintain continuity, key terminology is included in the "Glossary" section of the appendix. For those lacking a foundation in MySQL, please refer to the related content in the appendix or consult dedicated MySQL books [69].
 
 ## Special Terminology Explanation
 
diff --git a/References.md b/References.md
@@ -118,16 +118,23 @@
 
 [59] J. M. Hellerstein, M. Stonebraker, and J. R. Hamilton. Architecture of a database system. Foundations and Trends in Databases. 1(2) pp. 141--259, 2007.
 
-[60] https://github.com/session-replay-tools/tcpcopy.
-
-[61] <https://github.com/session-replay-tools/cetus>.
+[60] Chuanpeng Li, Chen Ding, and Kai Shen. Quantifying the cost of context switch. In Proceedings of the 2007 workshop on Experimental computer science, ExpCS ’07, New York, NY, USA, 2007. ACM.
+[61] Xiangpeng Hao, Xinjing Zhou, Xiangyao Yu, and Michael Stonebraker. 2024. Towards Buffer Management with Tiered Main Memory. Proc. ACM Manag. Data 2, 1 (Feb. 2024), Article 31. SIGMOD.
 
 [62] Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. 2012. Understanding and detecting real-world performance bugs. In ACM SIGPLAN Conference on Programming Language Design and Implementation, PLDI '12, Beijing, China - June 11 - 16, 2012. 77–88.
 
 [63] Jelena Antic, Georgios Chatzopoulos, Rachid Guerraoui, and Vasileios Trigonakis. 2016. Locking made easy. In Proceedings of the International Middleware Conference (Middleware). 1--14.
 
-[64] https://github.com/enhancedformysql/enhancedformysql.
+[64] D. Dice and A. Kogan, "Avoiding scalability collapse by restricting concurrency" in Euro-Par 2019: Parallel Processing, Cham:Springer International Publishing, pp. 363-376, 2019.
+
+[65] https://github.com/session-replay-tools/tcpcopy.
+
+[66] https://github.com/session-replay-tools/cetus.
+
+[67] https://github.com/enhancedformysql/enhancedformysql.
+
+[68] https://github.com/enhancedformysql/benchmarksql.
 
-[65] https://github.com/enhancedformysql/benchmarksql.
+[69] https://github.com/enhancedformysql/tech-explorer-hub.
 
 [Next](Appendix.md)
