Plans and Limits

In SPG99, it is useful to distinguish between two things:

• product plans and user-facing sizes;
• the actual runtime contract of the managed environment.

Trial hard limits

The trial scenario has strict limits:

• up to 1 tenant;
• up to 2 databases;
• up to 10 CU·h of compute;
• up to 5 GiB of egress.

In the UI, the trial plan is usually additionally limited to the L1 size.

PAYG

The product line already includes the standard_payg plan:

• without the hard trial limits on the number of resources;
• with compute, storage, archive, and egress billed by usage;
• with the user-facing size scale L1–L5.

In practice, this means the platform is already prepared for the PAYG model and for the full user size line.

What matters about sizes today

It is important to distinguish honestly between two levels.

1. The user-facing size model

From the product and UI point of view, the user sees the following sizes:

• L1
• L2
• L3
• L4
• L5

2. The active autoscale runtime contract

In the current production serverless handoff contract, the writer autoscaler works for:

• L1
• L2

In other words:

• the fully active autoscale handoff today is L1 <-> L2;
• the PAYG line and the L1–L5 size scale are already prepared on the product side.

What size means

size is a clear product label. It affects:

• the expected CPU and memory class;
• managed PostgreSQL parameters;
• connection limits;
• autoscaler and routing behavior.

It is important to understand:

• size is not a promise of “the same VM forever”;
• the actual mapping to CPU, memory, and managed settings is defined by the platform;
• during an autoscale handoff, the writer may move to a new compute generation.

What matters about scale

The public API still exposes the binary fields:

• current_scale
• target_scale

In the current contract, they mean only:

• 0 — the writer is not running;
• 1 — the writer is running.

For the writer profile, it is now more important to look at:

• current_profile
• target_profile
• scale_state

Practical meaning

• trial helps you start quickly and try the platform safely;
• PAYG already describes the production economic model;
• the product size line is broader than the currently active autoscale handoff;
• for today's production serverless operation, the main active range remains L1–L2.