MySQL Contributor Summit 2026の振り返りです。参加者登壇者にはAmazon、Google、Booking. com、Percona、ProxySQL、Readyset、Zhao Song氏（Hopsworks）、VillageSQL、MariaDB Foundation、MySQL ACE コントリビューターが含まれていました blogs.oracle.com/mysql-jp/my… #mysql_jp

Absolutely. The same backend connection can be used by multiple clients if it is safe to share it: no transactions, no temporary tables, no user defined variables, etc . ProxySQL tracks the statuses of frontend and backend connections.

Tell more about connection multiplexing please. Is it possible to use same connection between several client threads?

Exactly

That's exactly why I built @ProxySQL ProxySQL is a **high-performance reverse proxy** specifically designed for MySQL, MariaDB, Percona, TiDB, and PostgreSQL. It sits between your application and your database cluster and gives you: • Connection multiplexing & pooling • Read/write splitting • Query routing, rewriting & caching • Load balancing failover • Advanced security (query firewall, SSL, etc.) • Traffic shaping & mirroring All while completely hiding your actual database servers from the application layer. If you're running databases at any meaningful scale, a good database reverse proxy becomes essential. Happy to answer any ProxySQL questions!

the app holding a direct connection to a specific primary is the bug everyone ships first. a stable proxy endpoint means failover becomes the proxy's problem, not 14 services all needing a redeploy

ProxySQL’s query cache continues to be one of the highest-ROI features for read-heavy workloads. Enabled a well-tuned rule on a production cluster and saw repetitive SELECTs drop from multi-millisecond to sub-millisecond latency. Zero app changes. Configured simply via cache_ttl in mysql_query_rules, with full control over which queries get cached and for how long. If you're running MySQL at scale, definitely worth testing on your hot read paths. What's your experience with it? Hit ratios? #ProxySQL #MySQL #Performance

'Knows when not to write code.' Past senior, that is the job. Half my impact at ProxySQL has been deletion — removing features, simplifying configs, killing clever rewrites before they metastasize. The senior who actually earns the 2am pages is not the one with the cleverest fix. It is the one who notices the system does not need a fix at all.

OLTP at scale: MySQL. I build ProxySQL — most million-QPS systems I see are MySQL or Postgres. Postgres wins on features, Mongo on docs, SQLite for embedded. "Favorite" is the wrong question — pick the one that survives your 3am pages.

this will definitely benefit a lot for sure

glad you dropped it I don't think anybody knows this

One of @ProxySQL’s most underrated features: transparent query comments for observability, zero app code changes. You can automatically tag queries by service, user, feature flag, or environment and suddenly your monitoring (pg_stat_statements, performance_schema, Prometheus, etc.) becomes crystal clear. Debugging production traffic goes from painful to straightforward. Who else is using this? What tags have you found most useful? #ProxySQL #Observability #PostgreSQL #MySQL

I see your point, and I can relate. At @proxysql this is exactly the type of problems we solve.

With all the Postgres scaling conversations right now, one thing I keep seeing teams fight: fragile direct connections during failovers, upgrades, or scaling. @ProxySQL’s stable endpoint health-aware routing turns those chaotic backend changes into completely invisible events for your apps. No connection string updates. No app-side retry logic changes. This architectural decoupling is why most teams that adopt it never go back to raw DB connections. How critical is endpoint stability in your current setup? #ProxySQL #PostgreSQL #MySQL #HA

A takeaway from Percona Live 2026: MySQL and PostgreSQL are different engines. But many production challenges are similar: • connection management • routing • failover • observability • multi-cluster operations That is why the traffic/control layer matters. proxysql.com/blog/percona-li… #ProxySQL #PerconaLive

Percona Live 2026 showed how broad the open source database ecosystem has become. MySQL remains strong. PostgreSQL momentum is impossible to ignore. Operational challenges are increasingly shared across both. Our reflections: proxysql.com/blog/percona-li… #ProxySQL #PerconaLive #MySQL #PostgreSQL