Performance

This page covers server-side and client-side performance tuning for Inspire.

Server Performance

Database

Symptom	Cause	Solution
Slow page loads	Missing indexes	Check PostgreSQL query plans; indexes are created automatically but custom queries may need tuning
High CPU on database	Too many concurrent queries	Increase connection pool size or add read replicas
Growing disk usage	Snapshot retention too long	Reduce snapshot retention in DataPool settings

Edge Service

Symptom	Cause	Solution
Slow API responses	High load	Scale horizontally with multiple Edge containers
Memory growth	Large tenant data	Increase container memory limit; check for memory leaks in logs
Slow image uploads	Disk I/O bottleneck	Use SSD storage; consider object storage for media

DataPool Service

Symptom	Cause	Solution
Data push delays	Too many concurrent streams	Scale DataPool service horizontally
Redis memory full	Too many cached values	Increase Redis memory limit or reduce cache TTL
SignalR connection limits	Too many connected devices	Add more DataPool instances behind load balancer

Client Performance

Display Devices

Optimisation	Details
Use appropriate resolution	Do not render 4K content on 1080p displays
Optimise images	Resize images to display size before uploading
Limit concurrent videos	Each video stream uses significant bandwidth and CPU
Use components over raw elements	Components are pre-rendered and cached efficiently
Reduce DataPool subscriptions	Only bind to fields you actually display

Web UI

Optimisation	Details
Use a modern browser	Chrome or Edge latest for best performance
Close unused Designer tabs	Each open Designer tab maintains a live connection
Limit canvas elements	Scenes with 100+ elements may be slow to edit — consider using components

Scaling

Horizontal Scaling

Inspire services can be scaled horizontally:

Service	Scale Method	Considerations
Edge	Multiple containers behind load balancer	Session affinity required
DataPool	Multiple containers with shared Redis	SignalR backplane via Redis
Renderer	Client-side, no scaling needed	One renderer per device

Redis Clustering

For large deployments (500+ devices), configure Redis in cluster mode:

Set up a Redis cluster with 3+ nodes
Update REDIS_URL to point to the cluster
Restart Edge and DataPool services
Verify connections in service logs

Most installations (up to 200 devices) run comfortably on a single server with default settings. Only consider scaling when you observe sustained resource pressure.

Monitoring

Use the built-in admin dashboard (Settings > System Health) to monitor:

CPU, memory, and disk usage per service
Database query performance
Active WebSocket connections
Data throughput (items/second)
Error rates