Use flexible resources
Back to home
On this page
Sometimes you have apps that are more memory-intensive.
In such cases, you can adjust the amount of memory using the resources key in your app configuration.
All memory must stay within the limits set by your plan.
Note
If you don’t understand how your resources are being allocated or are having issues configuring them, open a support ticket. The Support team can look into the details of your project.
The resources key has the following possible options:
| Name | Type | Minimum | Description |
|---|---|---|---|
base_memory |
integer |
64 | The base amount of memory in MB to be given to the container. Up to 1024. |
memory_ratio |
integer |
128 | The amount of memory in MB that increases with CPU size. Up to 1024. |
The memory allocated to the container is calculated as the base memory plus the memory ratio multiplied by the CPU:
memory = base_memory + (memory_ratio * CPU).
When the resources key is set, the CPU sizes come from the following table:
| Size | CPU |
|---|---|
| XS | 0.25 |
| S | 0.5 |
| M | 1 |
| L | 2 |
| XL | 4 |
| 2X L | 8 |
So you might have the resources set as follows:
resources:
base_memory: 128
memory_ratio: 180If your app is set with a size of S, it gets 218 MB of memory: 128 + (0.5 * 180) = 218.
If you change the size to L, it gets 488 MB of memory: 128 + (2 * 180) = 488.
Other containers and workers 
Setting the resources key on one app doesn’t affect other apps in your project.
So you can set it for one app and have your other apps be assigned resources automatically.
Workers based on apps with the resources key default to inheriting the set resources.
To free memory for your main app,
set the values within the resources key to be lower for your workers.
Performance profiles
The two options for resources allow for different performance profiles.
Some apps, such as many PHP apps, don’t require so much memory to start
but need a lot of memory for each parallel request.
Such apps benefit from a low base_memory and a high memory_ratio.
Other apps, such as those with a persistent runtime including many Java and Go apps,
require more memory to get started
but don’t need much memory for each parallel request.
Such apps benefit from a high base_memory and a low memory_ratio.