Updated on 2026-11-17 GMT+08:00

Application Performance Specifications

RDS Supabase includes a connection pooling component called Supavisor. Designed specifically for RDS for PostgreSQL and released as open-source software, Supavisor is a cloud-native, distributed database connection pooler. It plays a critical role in the system as a traffic scheduling and connection orchestration layer. Two of its core capacity control parameters, POOLER_DEFAULT_POOL_SIZE (default backend connection pool size) and POOLER_MAX_CLIENT_CONN (maximum number of client connections), are also key parameters in the performance specifications presented to users. The two parameters control connection behavior at the backend (connections to the database) and frontend (connections to the clients), respectively.

Table 1 Connection pool capacity and performance parameters of RDS Supabase

Parameter

Description

Applies To

Primary Purpose

POOLER_DEFAULT_POOL_SIZE

Defines the maximum number of actual TCP connections that the Supavisor connection pool maintains with the underlying hosted RDS for PostgreSQL instance for each database user or tenant.

Supavisor ↔ PostgreSQL database

Protects the database against memory exhaustion.

  • Database protection mechanism: Each connection to the RDS for PostgreSQL instance consumes a significant amount of server memory (typically several to tens of megabytes). This parameter sets a hard limit to ensure that Supavisor never establishes more than the specified number of connections to the database, preventing the database from crashing due to excessive connections (out of memory, or OOM).
  • Connection reuse and queuing: When the number of concurrent frontend requests exceeds this limit, Supavisor does not request additional connections from the database. Instead, excess requests are queued in the pool until existing requests release their backend connections, which can then be reused.
  • Default behavior: In a multi-tenant environment, this is the default capacity of each individual connection pool.

POOLER_MAX_CLIENT_CONN

Defines the maximum total number of concurrent connections that Supavisor can accept from frontend clients, such as API services, edge functions, and frontend applications.

Application/Client ↔ Supavisor

Allows high concurrency and accommodates a large volume of application requests.

  • High-concurrency capacity: This parameter is central to connection pooling. Because client connections (from applications to Supavisor) are lightweight, Supavisor can handle far more client connections than the underlying database can support.

  • Traffic control: This parameter determines how many instances or processes can remain connected simultaneously. If an application deploys a large number of edge functions (such as functions provided by Huawei Cloud FunctionGraph), each function occupies a client connection. Once this limit is reached, new client connection attempts to Supavisor are rejected.

The default connection pool capacity and performance parameter values vary depending on the resource specification tier of RDS Supabase. For details about the default values and constraints for each tier, see Table 2.

Table 2 Connection pool capacity and performance parameters of RDS Supabase

Specification Scenario

Specification Code

Number of CPUs

Memory (GB)

Default Backend Connection Pool Size

Maximum Number of Client Connections

Specifications Required for the Managed RDS for PostgreSQL Instance

2 vCPUs | 4 GB single-node (free by default)

rds.supabase.n1.large.2

2

4-

20

100

  • ≥ 2 vCPUs | 4 GB
  • Primary/standby or single-node

2 vCPUs | 8 GB single-node (free by default)

rds.supabase.n1.large.4

2

8

20

500

  • ≥ 2 vCPUs | 8 GB
  • Primary/standby or single-node

4 vCPUs | 8 GB single-node (subject to evaluation)

rds.supabase.n1.xlarge.2

4

8

40

1000

  • ≥ 4 vCPUs | 8 GB
  • Primary/standby or single-node

4 vCPUs | 16 GB single-node (subject to evaluation)

rds.supabase.n1.xlarge.4

4

16

40

1200

  • ≥ 4 vCPUs | 16 GB
  • Primary/standby or single-node

8 vCPUs | 16 GB single-node (subject to evaluation)

rds.supabase.n1.2xlarge.2

8

16

80

2400

  • ≥ 8 vCPUs | 16 GB
  • Primary/standby or single-node

8 vCPUs | 32 GB single-node (subject to evaluation)

rds.supabase.n1.2xlarge.4

8

32

80

2800

  • ≥ 8 vCPUs | 32 GB
  • Primary/standby or single-node

16 vCPUs | 32 GB single-node (subject to evaluation)

rds.supabase.x1.4xlarge.2

16

23

160

5600

  • ≥ 16 vCPUs | 32 GB
  • Primary/standby or single-node

16 vCPUs | 64 GB single-node (subject to evaluation)

rds.supabase.x1.4xlarge.4

16

64

160

6400

  • ≥ 16 vCPUs | 64 GB
  • Primary/standby or single-node

When using RDS Supabase in production, ensure that the specifications of the PostgreSQL instance hosted by Supabase match the specifications of the RDS Supabase application. You should also control the number of client connections to avoid exceeding the maximum connection limit. Avoid excessive long-lived connections and frequent, time-consuming requests. Otherwise, connections may be rejected, resulting in workload timeouts or interruptions. In severe cases, all available PostgreSQL connections may be exhausted, leading to slow execution and response.

After an RDS Supabase application manages a DB instance, its internal components consume 40 to 50 database client connections by default. Therefore, use a DB instance with relatively low utilization (that is, one receiving fewer client requests) whenever possible. This helps prevent insufficient available connections from causing customer workload timeouts, interruptions, or even unavailability.