> ## Documentation Index
> Fetch the complete documentation index at: https://docs.presenton.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Performance and scaling

> Measure and tune self-hosted Presenton workloads.

Presentation generation combines provider calls, document processing, application work, storage, and browser-based export. Measure each stage separately before adding resources.

## Common constraints

| Symptom                          | Likely area to inspect                                |
| -------------------------------- | ----------------------------------------------------- |
| Slow outline or slide generation | Text-provider latency, model capacity, or prompt size |
| Slow document ingestion          | File size, OCR, parser workers, CPU, or memory        |
| Slow image generation            | Image-provider latency or local GPU capacity          |
| Export timeouts                  | Available memory, CPU, and internal rendering logs    |
| Degraded concurrent jobs         | Provider rate limits and host resource saturation     |

## Scaling guidance

* Establish a baseline with representative documents and slide counts.
* Prefer asynchronous API workflows for long jobs.
* Increase document parser workers only after measuring CPU and memory.
* Keep persistent storage fast enough for uploads, assets, and exports.
* Check provider rate limits before scaling the Presenton host.
* Test database migrations and shared storage before attempting multiple application instances.