The API server is the HTTP service that handles every API request. In most ActiveManage deployments it runs in the same PHP-FPM process pool as the web UI. For very high-load deployments you can split it onto dedicated infrastructure.
Default Setup
A single PHP-FPM pool serves both / (web UI) and /api (REST). Nginx routes by path. This is the simplest topology and fine up to roughly 500 requests/second.
Splitting the API onto a Dedicated Pool
- Provision a second PHP-FPM pool with its own worker count, memory limit and timeout.
- Update Nginx to send
location /apito the new pool. - Optionally use a separate FPM config with tighter PHP memory limits — API requests are usually smaller than full-page renders.
- Health-check both pools independently; the platform exposes
/healthzon each.
Why Split?
- Different resource profiles: Web requests render full pages; API requests are typically smaller and faster. Different worker counts make sense.
- Stability isolation: A surge of API traffic doesn't crowd out interactive users.
- Independent scaling: Add API capacity without doubling web capacity.
- Auditing and logging: Separate log files make analysis cleaner.
Worked Examples
- Small SaaS: Single combined pool, 16 workers, 2GB total memory. Sufficient until ~200 concurrent users.
- Growing mid-market: Split into web (24 workers) and api (16 workers). Each scaled independently per usage.
- Marketplace at scale: Dedicated API hosts, autoscaled by Kubernetes based on request rate. Web tier separately scaled by page-views.
