Marat Surmashev

A baseline note for our performance at scale series: what we measure first, why the starting point matters, and how we keep the benchmark honest.

We're benchmarking popular open-source FHIR servers and Aidbox across CRUD, batch processing, and search workloads. Here's what we're testing and why.

Benchmarking Aidbox with 8 CPU cores: single instance vs multiple. 2×4 loses only 6% RPS but adds redundancy and better P99.

Bug report → full test report in 10 minutes. See how LLM agents reproduce bugs, generate tests, and validate Aidbox automatically.

We load-tested Aidbox pool settings. 80 configs show how web threads and DB pools affect throughput and latency.

What if working with FHIR didn’t start with APIs and clients, but with files and folders? This article explores FHIR-FUSE, a filesystem interface for FHIR servers, what it makes easier, and where its limits become clear.

Learn how subscriptions simplify data management with automated, event-driven workflows. They improve care coordination, decision-making, and streamline processes like data syncing and automation. Plus, Aidbox offers reliable, scalable solutions for both simple and high-load systems.

Big data usually means big problems and a lot of work. One of the problems is how to load a lot of data. There is usually some source of big data, and the problem is posed as some kind of legacy system migration or migration of historical data into an FHIR server.