Optimizing data transfer costs with Egress Cost Optimizer¶

How Egress Cost Optimizer works¶

Egress Cost Optimizer analyzes your listing configuration in terms of the number of regions and cloud providers where the listing is available, and delivers the most cost-efficient auto-fulfillment. For example, if you’re replicating data to multiple cloud regions and incurring repeated egress costs on the same dataset, it intelligently routes the data through a Snowflake-managed ECO cache. In this way, customers end up paying zero additional egress costs to expand to new regions, reducing the data transfer costs.

In another example, if you’re only replicating to 1-2 regions within the same cloud provider, the ECO doesn’t use the ECO cache because your data transfer costs are already optimized. As a result, by turning on ECO, you’re ensuring minimum data transfer costs under any data sharing scenario. For more information on costs, benefits, and limits of ECO see Benefits and costs of egress cost optimization and Limitations of ECO.

ECO doesn’t impact existing security, features, and performance commitments of listings, such as support for data encryption in transit and rest through Snowflake Tri-Secret Secure (TSS), or existing cross-cloud auto fulfillment features (for example object-level replication, listing refresh cron schedule, and listing refresh history).

You can learn more about the Snowflake supported third-party sub-processors that are leveraged in connection with Cloud Cache by visiting our Sub-processor site.

When using ECO, your data will be hosted in the following regions, in addition to the regions where you make the data available to your consumers:

ECO ensures that under any circumstance, you’re only paying cross-cloud egress cost once. As a result, the more cloud regions that you replicate to, the more the potential egress cost savings.

Benefits and costs of egress cost optimization¶

Egress cost optimization can be used to reduce and control listing auto-fulfillment costs.

Initial costs:

The first time data is auto-fulfilled using the egress cost optimizer, the data is cached in Snowflake-managed S3-compatible storage with zero-egress costs, and you are charged for the initial egress of all the data in each listing to this storage location. Thereafter, egress is charged only for data updates.

Incremental data loading vs full data reloading:

If you regularly replace tables, or truncate and reload tables, be aware that this fresh data will be treated as a new table. Using these processes causes those tables to be re-cached, which incurs a higher cost than modifying the data by using less resource-intensive methods.

Greater savings with many regions or clouds:

Sharing data across more regions increases your savings on total egress costs. The more regions where data is shared, the greater the savings with the egress cost optimizer.

Database level, not listing level:

Where an auto-fulfillment schedule is set on the account level, rather than on the listing level, the egress cost optimizer will be enabled on all the listings that follow the account schedule. After the cost optimizer is enabled on a database, all subsequent auto-fulfillment involving that database will use it.

For more information about pricing for egress between source and target regions or clouds, see the Snowflake pricing guide.

Limitations of ECO¶

• Incremental data ingestion is required for the cloud cache to be fully used by the egress cost optimizer.

• The cloud cache is only used by the egress cost optimizer for refreshes made by auto-fulfillment.