4 ms·
No thanks. In 2026 I want HA and replication out of the box without the insanity.
by otabdeveloper4 8mo ago
No thanks. In 2026 I want HA and replication out of the box without the insanity.
- nrvn 8mo agoExactly my thoughts immediately after reading the word “just”. Also, PITR.
- eikenberry 8mo agoCame to say the same thing. Personally I'd only touch Postgres in a couple cases. 1. Downtime doesn't matter. 2. Paying someone else (eg. AWS) to manage redundancy and fail-over. It just feels crazy to me that Postgres still doesn't have a native HA story since I last battled with this well over a decade ago.
- groundzeros2015 8mo agoYou exceeded the step of maxing out the best server you can buy?
- throwaway7783 8mo agoHA is not about exceeding the limits of a server. Its about still serving traffic when that best server I bought goes offline (or has failed memory chip, or a disk or... ).
- groundzeros2015 8mo agoReplication?
- lima 8mo agoPostgres replication, even in synchronous mode, does not maintain its consistency guarantees during network partitions. It's not a CP system - I don't think it would actually pass a Jepsen test suite in a multi-node setup[1]. No amount of tooling can fix this without a consensus mechanism for transactions. Same with MySQL and many other "traditional" databases. It tends to work out because these failures are rare and you can get pretty close with external leader election and fencing, but Postgres is NOT easy (likely impossible) to operate as a CP system according to the CAP theorem. There are various attempts at fixing this (Yugabyte, Neon, Cockroach, TiDB, ...) which all come with various downsides. [1]: Someone tried it with Patroni and failed miserably, https://www.binwang.me/2024-12-02-PostgreSQL-High-Availability-Solutions-Part-1.html https://www.binwang.me/2024-12-02-PostgreSQL-High-Availabili...
- riku_iki 8mo agoso, what do you use instead?