Optimization

Use these practices to improve Turbo Server performance for memory-intensive workloads and high connection counts.

What You'll Learn

• How to precache assemblies to improve memory sharing
• How to tune request threads for higher concurrency
• Where to configure these optimizations

---

Memory

For applications with heavy memory requirements, use the precache command to enable Windows DLL and EXE sharing. This reduces startup time and memory consumption for users on the application server. Example for MATLAB:

# Verify the image has been pulled and cached locally
turbo images

# Run the precache command
turbo precache mathworks/matlab

After precaching, enable Enable Assembly Cache on the workspace application's General tab in the admin site.

Networking

Maximum Number of Request Threads

Turbo Server uses an Apache reverse proxy for APIs and the Portal site. The proxy sets a maximum number of request threads; when exceeded, incoming requests queue and can appear as hangs to end users. Each thread carries startup and runtime memory costs, so set this value within your server capacity. For larger concurrency needs, consider Load Balancing.

Configure request threads via the admin CLI:

Server.exe admin global request-threads 2000

See Command Line Management for CLI usage. The default is 2000 in Turbo Server 21.9 and later; earlier releases default to 150. This setting maps to the mpm_winnt ThreadsPerChild value.

Graphics

Graphic application routing support is coming soon.