Instance Types

Bruin Cloud runs all assets in individual, ephemeral environments, called "instances". These instances are managed by the Bruin Cloud platform, and provide serverless compute for all of your assets.

You can configure the instance type for each asset inside your asset definitions:

instance: "b1.small"

The following instance types are available in Bruin Cloud:

Instance Type	CPU	Memory
b1.pico	125m	256Mi
b1.nano	250m	512Mi
b1.small	500m	1500Mi
b1.medium	750m	2400Mi
b1.large	1	4Gi
b1.xlarge	2	6Gi
b1.2xlarge	2.5	8Gi
b1.3xlarge	3	12Gi
b1.4xlarge	3.5	16Gi
b1.5xlarge	4	24Gi
b1.6xlarge	4.5	32Gi
b1.7xlarge	5	48Gi
b1.8xlarge	5	64Gi

If you don't set instance, Bruin Cloud picks one based on the asset type:

Asset Type	Default Instance
Python	b1.nano (b1.small when materialized as a table)
Ingestr	b1.medium
SQL	b1.nano
Sensors	b1.pico

Notes:

• CPU is measured in cores (1000m = 1 core)
• Memory is measured in gibibytes (Gi) or mebibytes (Mi)
• Values are the maximum each instance can use; on b1.xlarge and larger, CPU may be lower when the cluster is busy
• An asset that goes over its max memory is stopped with an out-of-memory error; pick a larger instance if this happens

How Instance Size Affects Concurrency

Larger instances consume more of your tenant's resource pool. See Concurrency & Resource Limits for details on the weighted-slot model.

Custom Instance Types

Bruin Cloud supports custom instance types. You can specify a custom instance type by setting the instance field to the name of the instance type you want to use. Talk to your account manager to get access to custom instance types.

Related

• Pipelines for how scheduled and manual runs use these instances.
• Asset definition for setting the instance field on an asset.