all posts
blog · topic

Ephemeral databases and Postgres

A database that exists for one test run, one pull request, or one agent task removes a whole class of accidents. These posts cover ephemeral and per-branch Postgres, cloning and point-in-time restore, database isolation for multi-tenant and agent workloads, and the operational side of running Postgres in its own microVM.

Ephemeral Postgres databases on PandaStack

86 posts

·11 min read

Leader Election When Your Compute Can Be Cloned

A restarted process comes up empty and has to ask who the leader is. A restored one never asks — it resumes mid-sentence still believing it holds the lock. Leases, fencing tokens, and why a forked sandbox must revoke leadership before it does anything else.

distributed-systemsleader-electionpostgres
Ajay Kumar
·11 min read

What "persistent" actually means in a sandbox

You wrote the file. You ran cat and saw it. Neither of those facts says the bytes are on a disk. Here is every layer a write passes through inside a microVM, which of them a crash erases, and why the honest answer for an ephemeral rootfs is not "fsync harder" but "get the artifact out".

durabilitystoragesandboxes
Ajay Kumar
·11 min read

The Best Prefect Hosting Platforms in 2026

Almost every 'how do I host Prefect' thread is two questions wearing one coat: where does the server live, and where do flow runs actually execute? They have different answers, and the second one is the one that costs money.

comparisonprefectorchestration
Ajay Kumar
·11 min read

The best ways to host MLflow in 2026

MLflow looks like a web app and is actually a database, a bucket, and a code-distribution channel wearing one name. An honest look at where to put each part — including where we're the wrong answer.

mlflowmlopsself-hosting
Ajay Kumar
·13 min read

Ephemeral Databases for AI Agents

Give an agent a run_sql tool and it will use it — including the DELETE it emits while debugging its own step. The fix isn't a better statement filter; it's a database that exists for one task and is deleted when the task ends.

ai-agentsdatabasepostgres
Ajay Kumar
·11 min read

The Best Dagster Hosting Platforms in 2026

You do not deploy 'Dagster'. You deploy a webserver, a daemon that must be a singleton, one or more code-location servers, and a Postgres you cannot skip. Here is how the hosting options handle that, honestly.

comparisondagsterdata-engineering
Ajay Kumar
·13 min read

Webhook Replay and Relay: Ingest First, Process Later

The provider retried for eleven hours, gave up, and the event is gone with no record you can inspect. The fix is boring and it is one afternoon of work: ingest first, process later.

webhooksarchitecturereliability
Ajay Kumar
·10 min read

Per-Tenant SQLite: One File, One microVM, and No Noisy Neighbours

One SQLite file per customer gets the data model right: blast radius is a path, erasure is rm. Then you put a thousand of them in one process and discover that a file boundary is not a failure boundary.

sqlitemulti-tenancymicrovm
Ajay Kumar
·10 min read

The best Meilisearch hosting platforms in 2026

Meilisearch is one stateful Rust binary with a memory-mapped index on a disk. That sentence explains every hosting option below, including the one where you do not host it at all.

comparisonmeilisearchsearch
Ajay Kumar

More posts in this topic