Performance Mode
Performance Mode lets you allocate more computing resources to a site running on a dedicated server. Enabling this mode increases the site’s workload priority, allowing Frappe Cloud to assign a larger share of web and background workers based on the overall workload of the server.
Availability
Performance Mode is available only for sites hosted on Dedicated Servers. Shared instances do not support this feature.
Enable or Disable Performance Mode
- Open the desired Site from your Frappe Cloud dashboard.
- Scroll down to the Dangerous Actions section.
- Locate the Performance Mode option.
- Toggle the setting to enable or disable High Performance mode.
Once the setting is changed, Frappe Cloud automatically recalculates worker allocation for all active benches on the server.
How Worker Allocation Works
Worker allocation is handled automatically using the server’s available resources. Every active Bench receives a proportional share of the server’s worker capacity based on its overall workload.
The allocation process is managed by Frappe Cloud’s automatic worker scaling mechanism, which distributes both web workers and background workers according to workload.
Workload Score
Each Bench receives a workload score calculated from all active, pending, or updating sites hosted on that Bench.
Performance Mode directly affects this score.
| Performance Mode | CPU Time Per Day | Priority |
|---|---|---|
| High Performance | 86,000 | Higher worker allocation |
| Standard | 8,000 | Normal worker allocation |
A higher workload score increases the Bench’s share of available workers on the server.
Server Memory Allocation
Before allocating workers, Frappe Cloud determines how much server memory is available for application processes.
- Approximately 60% of usable memory is allocated for Gunicorn (web) workers.
- Approximately 40% is reserved for Background Workers.
This ensures that both web requests and background jobs have sufficient resources while maintaining overall server stability.
Per-Bench Worker Allocation
Each Bench receives workers proportional to its workload compared to the total workload on the server.
Worker allocation is automatically limited within predefined minimum and maximum values.
| Worker Type | Default Minimum | Default Maximum |
|---|---|---|
| Gunicorn (Web Workers) | 2 | 36 |
| Background Workers | 1 | 8 |
If custom worker limits are configured for the Release Group, those values override the defaults.
Threaded Workers
If threaded Gunicorn workers are enabled for a Bench, Frappe Cloud reduces the total number of Gunicorn processes while maintaining approximately the same request-handling capacity.
This helps reduce memory usage without significantly affecting concurrency.
Memory Limits
When Bench memory limits are enabled, Frappe Cloud automatically calculates memory limits for every Bench based on the number of allocated workers.
The platform configures:
- Memory High limit
- Maximum Memory limit
- Swap Memory limit
If memory limits are disabled, these restrictions are removed and no per-Bench memory limits are applied.
Best Practice
Enable High Performance mode only for sites that consistently require additional computing resources. Since worker allocation is shared across all active Benches on a server, prioritizing one workload may reduce the resources available to others.
