Fair Usage Policy
Learn about how your compute entitlement works.
Observe by Snowflake's ingest subscription plans include the compute you need to query, transform, monitor, and explore your data, and, with an AI package, the AI capabilities to investigate it. Compute is measured in Observe Compute Credits (OCC) and AI usage is measured in Observe AI Credits (OAIC). Together they make up your Observe Platform Unit (OPU) entitlement.
The Fair Usage Policy explains how those entitlements work, what happens if your usage approaches or exceeds them, and the tools available to help you stay in control. The policy is designed to do one thing: keep the platform fast and predictable for you and every other customer. Most customers operate well within their entitlement and never need to think about it.
NoteThe Fair Usage Policy applies to ingest subscription plans. If you are on a consumption-based (usage) plan, you are billed directly for the ingest, OCC, storage, and OAIC you consume, and these entitlement limits do not apply.
Your compute entitlement
Every ingest subscription includes a compute entitlement proportional to your contracted ingest volume, measured in OCC. OCC are consumed by queries, Dataset acceleration, Monitors, data loading, and related platform operations. For the full calculation, see Compute and AI entitlement calculation.
Your AI entitlement
If your subscription includes an AI package, you have access to Observe by Snowflake's AI capabilities as set out in your subscription order form. AI features consume OAIC each time they call a model, based on the tokens used and the model's published rate. For which features use OAIC, see AI features and credit usage.
Your AI package includes a monthly OAIC entitlement based on the price of your AI package. For the calculation, see Compute and AI entitlement calculation.
One entitlement for compute and AI
Rather than hold you to a separate limits for compute and AI, Observe by Snowflake combines them. Your subscription includes one OPU entitlement that both compute and AI draw down. If you are light on compute, the difference is available for AI. If you are light on AI, the difference is available for compute. You don't have to decide the mix up front, and you can still see exactly where your consumption is going.
How usage is measured
Your OPU entitlement is a monthly allocation. Observe by Snowflake evaluates your consumption against it over time rather than at any single moment, so a short spike during an incident, a heavy analysis session, or a deep AI investigation will not, on its own, put you over your entitlement. What matters is sustained consumption.
See Compute and AI entitlement calculation for details about how OPU is calculated.
What happens as you approach your entitlement
Observe by Snowflake is designed to engage early and help you optimize before you ever experience a slowdown. As your sustained consumption rises, you can expect the following:
| Usage | What you can expect |
|---|---|
| Within your entitlement | Normal operation. No action needed. |
| Approaching your entitlement | A notification to your account administrators, along with guidance on your highest-consuming queries, Datasets, Monitors, and AI usage. If you have Observe by Snowflake Enterprise Support, your Technical Account Manager can provide a hands-on optimization review. |
| At or above your entitlement | Continued optimization support, and a conversation about right-sizing your plan if your needs have grown. Because compute and AI share one allocation, this may be as simple as rebalancing between them. |
| Sustained, significant overage | Observe by Snowflake may manage compute to protect platform performance (see below). Your account team will work with you on a plan to expand your entitlement or adjust usage. |
The goal at every stage is to help you get more out of the platform, not to cut you off.
How Observe by Snowflake protects platform performance
Observe by Snowflake is a multi-tenant platform. To keep performance consistent for everyone, Observe by Snowflake may manage compute when sustained consumption significantly exceeds an entitlement. Observe by Snowflake never blocks or drops your incoming data; ingestion always continues. Instead, compute-intensive operations may be slowed: queries may run on smaller warehouses, Dataset processing intervals may relax, and Monitor evaluation intervals may extend. In practice this feels like reduced performance on heavy workloads rather than a hard stop. Because compute and AI share one OPU entitlement, this applies however your consumption is split between them. Observe by Snowflake never disables AI capabilities, and nothing in your account is turned off.
Tools to manage your consumption
You have full, self-service control over your consumption. The following tools help you understand and tune usage:
| Tool | What it helps you do |
|---|---|
| License Dashboard | View your compute allocation against your entitlement. Ingest-based licenses only. |
| Usage Dashboard | Identify which Datasets, users, queries, and Monitors consume the most OCC. |
| AI Model Usage Dashboard | See AI Credit consumption by user, feature, and model, with a token breakdown. |
| Credit Manager | Set global, per-user, and per-dashboard OCC quotas for queries and acceleration. |
| AI Credit limits | Set per-user and account-wide AI Credit limits to protect against unexpected consumption. [EDITOR: link to the Terraform settings page once documented] |
| Acceleration Manager | View and adjust Dataset processing intervals to control acceleration cost. |
| Usage attribution | Break down consumption by team, application, or use case. |
For common ways to reduce compute consumption, such as right-sizing Monitors, loosening processing intervals on non-critical Datasets, and identifying high-cost queries, see the OPAL performance cookbook. For the per-workload limits that protect your account, see Platform limits and governors.
NoteOPU reporting, showing your combined consumption against your OPU entitlement with the OCC and OAIC breakdown underneath, is being added to the product. Until then, you can use the dashboards mentioned in this section, or have your account team can help you review your combined usage.
Updated about 1 hour ago