PerconaBytes3d ago
"70% of RAM" isn't tuning. It's a starting guess.
A 99% hit ratio doesn't mean healthy — you can be IO-bound at 99%. Watch free buffers, dirty page %, and the rate of change on Innodb_buffer_pool_reads.
https://percona.community/blog/2026/04/02/innodb-buffer-pool-tuning-from-rule-of-thumb-to-real-signals/ - LC
#MySQL #InnoDB
PerconaBytes5d ago
innodb_redo_log_capacity set to a default years ago and never touched? You're guessing.
Watch Innodb_log_waits. Measure LSN growth. Size for 30–60 min of peak redo, not a % of RAM.
https://percona.community/blog/2026/05/02/innodb-redo-log-sizing-stop-guessing-start-measuring/ - LC
#MySQL #InnoDB
InnoDB Redo Log Sizing: Stop Guessing, Start Measuring

Introduction Many MySQL configurations inherit redo log sizing from defaults, aging blog posts, or configuration folklore. innodb_redo_log_capacity gets set once… and then quietly fades into the background. But redo log capacity directly shapes how efficiently MySQL absorbs writes, manages checkpoint pressure, and handles burst-heavy workloads. Set it too low, and aggressive flushing can throttle throughput. Set it too high, and crash recovery can become painfully long. Redo logs are more than crash insurance.

Gea-Suan LinMay 23

https://blog.gslin.org/archives/2026/05/23/13044/innodb-%e7%9a%84-repeatable-read-%e8%88%87-next-key-lock-%e7%9a%84%e9%97%9c%e4%bf%82/

InnoDB 的 REPEATABLE READ 與 next-key lock 的關係

#db #innodb #key #lock #mysql #next #performance #phantom #query #range #rdbms #read #repeatable #row #speed #sql #transaction

InnoDB 的 REPEATABLE READ 與 next-key lock 的關係

最近遇到 InnoDB 效能的問題發現的，記錄起來。 在 SQL-92 裡面直接有提到 REPEATABLE-READ 允許 phantom read，也就是 search condition 不保證會會傳回一樣的 row(s)：(在文件上可以查到 REPEATABLE READ 對 P3 是 Possible) 3) P3 (Phantom): SQL-transaction T...

Gea-Suan Lin's BLOG
ikristinaApr 27

How MVCC Works and Why Databases Use It

https://ikristina.github.io/blog/mvcc

#databases #mvcc #postgres #concurrency #cockroachdb #mysql #InnoDB #transaction

How MVCC Works and Why Databases Use It

Multi-version concurrency control lets readers and writers proceed without blocking each other. A breakdown of the core mechanics, and how PostgreSQL, CockroachDB, and MySQL InnoDB implement them.

Threads of Thought
HabrMar 20

[Перевод] MariaDB 12.3: binlog внутри InnoDB

Коротко для ленивых В MariaDB 12.3 binlog можно хранить внутри InnoDB через binlog_storage_engine=innodb . Главный эффект: вместо двух fsync() на commit остаётся один, поэтому на write-heavy нагрузке резко растут TPS и снижается tail latency. В тестах из статьи прирост на полном durability-профиле составил примерно 2.4x–3.3x . Backup, restore и ресинк реплик становятся проще, потому что binlog и данные теперь консистентны на уровне одного механизма хранения. Цена за это: обязателен GTID, Galera пока не поддерживается, а innodb_log_file_size нужно подбирать внимательнее из-за роста объёма redo. Если у вас обычная схема primary + async replica на InnoDB, эту возможность точно стоит хотя бы протестировать.

https://habr.com/ru/articles/1011298/

#MariaDB_123 #InnoDB #binlog #GTID #репликация #производительность_SQL #crash_recovery #fsync #OLTP

MariaDB 12.3: binlog внутри InnoDB

Полноформатная адаптация для Habr по мотивам статьи Adrien Obernesser «MariaDB 12.3 – Binlog Inside InnoDB». Коротко для ленивых В MariaDB 12.3 binlog можно хранить внутри InnoDB через...

Хабр
Gea-Suan LinFeb 21

https://blog.gslin.org/archives/2026/02/21/12901/mariadb-12-3-%e5%b0%8d-innodb-%e7%9a%84%e5%a4%a7%e6%94%b9%e9%80%b2/

MariaDB 12.3 對 InnoDB 的大改進

#binlog #cpu #database #db #engine #fsync #i/o #innodb #latency #mariadb #mysql #performance #rdbms #speed #storage #write

MariaDB 12.3 對 InnoDB 的大改進

看到 Mark Callaghan 寫的「MariaDB innovation: binlog_storage_engine」、「MariaDB innovation: binlog_storage_engine, small server, Insert Benchmark」這篇，裡面提到了 MariaDB 12.

Gea-Suan Lin's BLOG
FromDual GmbHFeb 9

In the tests in my previous articles, I found that #MariaDB completely ruined the concept of the dynamically configurable #InnoDB buffer pool!

And this within a long-term support release series. To prevent this from affecting your production environment, I have described more about this here:

https://www.fromdual.com/blog/mariadb-dynamically-configurable-buffer-pool-broken/

Image Author: Cplakidas, GNU FDL v1.2

ShinguzFeb 9

Bei den Tests in meinen vorherigen Artikeln habe ich festgestellt, dass #MariaDB das Konzept des dynamisch konfigurierbaren #InnoDB Buffer Pools völlig kaputt gemacht hat!

Und dies innerhalb einer Long-Term-Support Release-Reihe. Mehr dazu, damit Euch das nicht in der Produktion auf die Füsse fällt, habe ich hier beschrieben:

https://www.fromdual.com/de/blog/mariadb-dynamisch-konfigurierbare-buffer-pool-kaputt/

Image Author: Cplakidas, GNU FDL v1.2

MariaDB FoundationFeb 6
New binlog implementation in MariaDB 12.3. Vastly improved performance by default, and crash-safe for those who had performance configs. https://mariadb.org/new-binlog-implementation-in-mariadb-12-3/ #mariadb #opensource #database #innodb
New binlog implementation in MariaDB 12.3 - MariaDB.org

I have recently completed a large project to implement a new improved binlog format for MariaDB. The result will be available shortly in the upcoming MariaDB 12.3.1 release. … Continue reading "New binlog implementation in MariaDB 12.3"

MariaDB.org
ShinguzFeb 4
At least some of the senior #PostgreSQL experts have identified the weak spots: "What's Missing in Postgres?" by Bruce Momjian: https://momjian.us/main/writings/pgsql/missing.pdf
That is a good starting point for improvement.
IMHO their outdated MVCC implementation is still missing from this presentation. And #InnoDB index clustered tables are really cool!