.. _use_binaries: ================================= Using Firely Server with Binaries ================================= Firely Server can be deployed in a non-cloud, locally hosted environment using native binaries. Afterwards it is possible to run Firely Server as described in the :ref:`Basic installation ` section. The necessary files can be downloaded as zip files. All versions (incl. historic versions) can be accessed `here `_. Please consider running a :ref:`reverse proxy ` when running Firely Server natively. Firely Server depends on the .NET Core platform and is therefore cross-platform in all supported `environments `_. For a production usage, Microsoft SQL Server or MongoDB need to be installed in the same environment in addition to Firely Server. Firely Server should be registered as a startup / system service (Windows Service or systemd on Linux) for operational reliability. Minimal platform requirements ----------------------------- Firely Server is a high-performance FHIR server designed for scalability and reliability. While it can be deployed in various environments, optimal performance depends on proper provisioning of resources. **Deployment Options & Operating System:** Firely Server is supported on all platforms supported by the `.NET framework `_. In practice, the choice of operating system should align with your team's operational expertise and familiarity with the platform. There are no limitations regarding any hypervisor being used when using a virtual machine instead of a physical server. **Memory (RAM)** - Minimum: 2 GB - Recommended: 16 GB or more, particularly when working with large datasets. Firely Server caches definitional FHIR artifacts (e.g., StructureDefinitions, ValueSets) in memory to enhance performance. **CPU** - Minimum: 2-core X86-64 CPU - Recommended: at least 4-core X86-64 CPU - Note: More cores benefit concurrent request handling **Disk Space** - Minimum: 1.5 GB of free disk space for the raw installation of Firely Server - Recommended: The recommendation depends on the selected production database and expected numbers of resources. MongoDB uses ~4GB for 1 million resources. SQL Server uses ~34GB for 1 million resources. - Note: SSD storage is strongly recommended for optimal I/O performance. See :ref:`vonk_performance` for additional considerations around choosing the right amount of system resources when scaling the server (i.e. available memory, CPUs).