Scale the deployment

The scale: section of the deployment manifest allows allocating additional computing resources for the squid add-ons and services. This option is only available for paid squids deployed in professional organizations.

info

Visit the Pricing page and/or our costs calculator if you're looking for an estimate.

The manifest supports the following scaling options:

dedicated:

Default: dedicated: false.

By default, the service deployments are collocated - that is, share resources with other deployments - and so the allocation of full resources is not guaranteed. With dedicated: true, the resources are reserved in advance.

We recommend setting dedicated: true for all squids running in production.

Squid deployment must be dedicated to be a subject of SQD Cloud SLA.

addons:

postgres:

See Postgres add-on for details.

services:

api:

Name	Description	Type	Default value	Optional
profile	Allocated resources profile	small | medium | large	small	Optional
replicas	The number of gateway replicas. The API requests are distributed between the replicas in the round-robin fashion	Number	1	Optional

The profile specifications for API service replicas are as follows:

Profile	colocated vCPU (max)	colocated RAM (max)	dedicated vCPU (requested)	dedicated RAM (max)
small	0.2	768Mi	0.5	768Mi
medium	0.5	1.5Gi	1	1.5Gi
large	1	3Gi	2	3Gi

processor:

Name	Description	Type	Default value	Optional
profile	Allocated resources profile	small | medium | large	small	Optional

The profile specifications for a processor service are as follows:

Profile	colocated vCPU (max)	colocated RAM (max)	dedicated vCPU (requested)	dedicated RAM (max)
small	0.2	768Mi	0.5	768Mi
medium	0.5	1.5Gi	1	1.5Gi
large	1	3Gi	2	3Gi

Example

squid.yaml

manifest_version: subsquid.io/v0.1
name: sample-squid

build: 

deploy:
  addons:
    postgres: 
  processor:
    cmd: [ "sqd", "process:prod" ]
  api:
    cmd: [ "sqd", "serve:prod" ]

scale:
  dedicated: true
  addons:
    postgres:
      storage: 100G
      profile: medium
  processor:
    profile: medium
  api:
    profile: large
    # load-balance three replicas
    replicas: 3