FHIR server and database

Aidbox is a FHIR-first platform combining a FHIR server and a purpose-built FHIR database under one system. Built on PostgreSQL with custom extensions, Aidbox delivers:

· High-throughput ingestion (up to 20,000+ resources/second)
· Low-latency search across millions of FHIR records
· Fast data export for analytics, migrations, or external system integrations
· Full control over indexing and data structure, letting you tailor performance optimizations and workflows to your exact needs

Unlike many solutions that abstract away the backend, Aidbox lets you access and manipulate your data directly — you can build custom business logic, implement advanced workflows, or tie into existing systems on your terms.

‍Health Samurai has implemented and supported dozens of high-load systems in production — spanning hospitals, labs, startups, healthcare software solution providers, and major payers — confirming the platform's performance and reliability.

1. FHIR-Native Database
Aidbox doesn’t just provide a FHIR API; it includes a FHIR-optimized database built on PostgreSQL with custom extensions. By combining the API and database in one platform, Aidbox ensures low latency, fewer moving parts, and less overhead than solutions that rely on multiple external services.

2. High-Performance Data Processing
Aidbox easily handles thousands of operations per second, even when working with millions of resources. This includes fast reads and writes, so you can load patient data, run analytics, or power clinical apps without delay.

3. Flexible Infrastructure
Because Aidbox is distributed as a lightweight Docker container that only needs PostgreSQL, you avoid the hassle of complex multi-service setups. It’s deployable on-premises, in the cloud, or in hybrid environments, all while reducing maintenance and potential points of failure.

4. Customizable Indexing & Data Model
Aidbox’s PostgreSQL foundation means you can define your own indexing strategies, modify search behavior, and shape the data model to match your specific project needs, rather than working around a one-size-fits-all approach.

1. High-volume data ingestion
We tested Aidbox with 23 million and then 100 million resources via the /fhir/$import endpoint.

• Single Import (23M Resources)
   • ~4.3 GB zipped (~17.9 GB unzipped)
   • Completed in ~65 minutes
   • ~6,000 resources/second

• Concurrent Import (100M Resources)
   • 4 concurrent requests, 10 parallel workers each
   • ~180 GB in the database post-import
   • ~20,000+ resources/second peak throughput

2. Fast bulk export
When exporting 100 million resources via /fhir/$export:

• 202.8 GB of data
• 1 hour 47 minutes total time
• 15,500 resources/second (32.3 MB/s)

3. High-concurrency CRUD operations
Even under 300 parallel threads, Aidbox achieves:

• Create: Up to 2,800 RPS (Observations)
• Read: Up to 3,500 RPS (Patients)
• Update: ~1,900 RPS (Encounters)
• Delete: Over 3,900 RPS (ExplanationOfBenefit)

4. Efficient transactions for mid-sized loads
Testing FHIR transactions (bundles of 10, 50, 100, 300 resources) showed:

• ~3,500 resources/second throughput
• Optimal concurrency at 1–2 threads per CPU core

5. Optimized search with indexing at scale
Search performance was validated across resources like Patient, Encounter, Claim, and Observation, with up to
5 million records:
‍
• Patient Search: ~2,081 RPS (search by name & sort by family)
• Encounter Search: ~3,125 RPS (by patient & date)
• Observation Search: ~2,042 RPS (by patient & date)

Flexible deployment
• Runs as a lightweight Docker container with a PostgreSQL backend
• No need for external search engines, queues, or middleware
• Works seamlessly in cloud, on-premises, or hybrid environments

Ideal For
• Healthtech teams building scalable digital health products
• Payers and providers that need fast, real-time access to clinical data
• Organizations working with real-time EHR feeds or running analytics on FHIR data

Aidbox offers speed, control, and reliability in one streamlined package. The result is a solution that can scale from tens of millions to billions of FHIR resources, delivering:Outstanding performance on minimal hardware
Even an 8 vCPU, 8 GB RAM setup can yield results 10x faster than popular open-source FHIR servers.

1. Streamlined operations
Aidbox’s containerized distribution requires only PostgreSQL, minimizing deployment and maintenance.

2. Reliability under heavy load
High concurrency and large transaction bundles don’t hamper Aidbox’s responsiveness or stability.

3. Scalable Bulk Import & Export
Process large FHIR datasets—from millions to billions—end-to-end, quickly and without added complexity.

If you’re looking for a powerful, FHIR-compliant platform that simplifies infrastructure, accelerates performance, and keeps data fully accessible, Aidbox is ready to meet your needs.