Performance tuning
TurboStack ships with sensible, server-aware defaults so most sites are fast by default. This page explains the caching layers available to you and how to override the defaults safely.
Faster pages improve conversion and search ranking, and they keep your site responsive during traffic peaks such as sales campaigns. For online stores, caching is the main tool for staying fast under load.
Important
TurboStack auto-tunes memory sizing to the size of your server. Only override these values with measured evidence, such as memory pressure, a low cache hit ratio, or garbage-collection (GC) pressure. Guessing at larger values usually makes performance worse, not better.
Caching layers
Caching is the highest-impact lever for most sites. TurboStack provides several layers, each suited to a different workload.
Redis
The cache instance on port 6379 holds transient data and is safe to clear. The persistent
instance on port 6378 stores durable data such as sessions and queues, so clearing it can log
users out or drop queued jobs. Configure Redis from the host's
Services tab.
Varnish
Varnish is a full-page cache that sits in front of PHP storefronts. It is most effective for read-heavy catalog and content pages on Magento, Shopware, and WordPress.
OPcache
OPcache stores compiled PHP bytecode so scripts do not have to be recompiled on every request. You can tune it per application under Applications > Technologies > PHP advanced options, where you can enhance OPcache and set a preload script.
Sizing keys you can override
These values are auto-tuned to your server by default. Change them only with evidence.
PHP-FPM and OPcache
PHP performance is tuned per application. Under
Applications > Technologies, the PHP advanced settings let you
adjust the PHP-FPM process manager (pm start, min spare, and max spare servers) and enable OPcache
preloading. Raise the PHP-FPM worker counts only when you see requests queuing and have the memory
headroom to support them. Each worker is one PHP process handling one request at a time.
Measure first
Always find the real bottleneck before changing anything. A slow page is often caused by a single layer (database, cache, or PHP) and tuning the wrong one wastes effort.
- Check the host's Health tab for resource pressure.
- Use Monitoring to inspect trends over time.
- See Monitoring concepts to understand what the metrics mean.
Once you have identified the constrained resource, apply the smallest override that relieves it and re-measure.
Related
- Services
- Applications
- Health
- Monitoring concepts
- TurboStack CLI - clear Redis, Varnish and OPcache on the server