Organizations increasingly rely on Snowflake’s Data Cloud to connect their business to the data they need, when they need it. Ensuring this data is always available, accessible, and accurate to all users and applications is critical. This means organizations need to trust that their data and platform are resilient, even in the event of an outage or natural disaster. Using Snowflake Database Replication and Failover/Failback makes this simple, so organizations such as CarVal Investors and Zeta Global can ensure business continuity across regions and across clouds. In just the last 12 months, Snowflake customers have replicated over 180PB of compressed data across regions and clouds to keep their businesses up and running. 

Today, we are excited to further advance these capabilities with Snowflake Client Redirect, which is now available in public preview. Client Redirect enables seamless redirection of client connections across Snowflake accounts in different regions and clouds for business continuity and minimal disruption to end users. Together with Snowflake Database Replication and Failover/Failback, users are able to recover data and client connections in seconds, at virtually any scale. For even higher resiliency, these features can even be used to perform failover across public clouds.

“Business continuity is a key requirement for our business. The new client redirect functionality in Snowflake allows us to continue to innovate while meeting our mission-critical requirements.” 

—Bharat Goyal, Senior Vice President, Engineering, Zeta Global

SNOWFLAKE’S UNIQUE APPROACH TO BUSINESS CONTINUITY

To maintain business continuity, organizations previously were faced with trade-offs. They could choose replication-failover solutions, which minimized data loss and recovery, but that would double their costs on both infrastructure and licensing. Or, they could choose backup-restore solutions, which were more cost-effective but which risked data loss and meant recovery time could take hours or even days depending on the data volumes. 

No longer does that need to be the case. Snowflake’s unique platform architecture ensures minimal data loss and instant recovery, all at a fraction of the cost. With Snowflake’s replication and failover capabilities, highly compressed micro-partitions are replicated to any region or cloud of choice, improving both storage efficiency and data freshness at the replication site. Paired with the ability to spin up compute resources instantly anywhere in the world, organizations are able to recover in seconds while paying only for the compute resources they need at the time. 

WHY USE SNOWFLAKE CLIENT REDIRECT?

During an outage, recovery and failover are not just limited to data. When the region hosting a primary deployment goes down, this disrupts all the client applications running against that data, such as business intelligence dashboards and data engineering pipelines. 

In the past, to recover, the connection string in every client application had to be manually changed to connect to the new primary site (the former secondary site) in a different region or cloud. Most organizations have hundreds or even thousands of client applications, so this process could take hours and prolong the recovery of mission-critical reports. Administrators would have to coordinate across several teams to make the changes, and tasks often fell on individual analysts and data scientists, which could lead to errors and added delays. Administrators struggled to ensure expedient recovery of applications, and business teams would lose access for hours to the business insights they relied on.

With Client Redirect, administrators can recover applications in seconds, even across a large number of  users, teams, and tools. Using a single server-side command, administrators can redirect client applications to the new primary site, without disrupting business teams. This shortens the recovery time for mission-critical workloads, eliminates the need for broad coordination across teams, and prevents error-prone changes from individual end users. 

“We chose Snowflake for its innovative approach to data and analytics. Client Redirect is just another example of how Snowflake continues to innovate and opens up new opportunities for us to use the platform for more workloads.”

—Albert Obeng, Director of Technology, CarVal Investors

However, it’s not enough to just have the capabilities to fail over in the event of an outage. Any downtime can result in lost revenue, liabilities, and more. Organizations need to trust that their teams know the protocols and everything will work when needed. Snowflake’s replication and failover capabilities, paired with Client Redirect, allow teams to perform disaster recovery drills so they can best prepare for a recovery during an outage while minimizing the impact to the business. 

HOW IT WORKS

To get started, you use a CONNECTION, a new Snowflake object that represents a URL that can be failed over across accounts. You create a primary connection in your Snowflake account and one or more linked secondary connections in accounts in other regions. Here’s an example: 

CREATE CONNECTION <connection_name>

Behind the scenes, Snowflake creates a new URL for this connection: 

https://<organization_name>-<connection_name>.snowflakecomputing.com

Now you can use this Connection URL in client applications such as your business intelligence dashboards and reports. This URL points to the account with the primary connection. During an outage in the primary region, you can promote a linked secondary connection in another account to serve as the primary, for example:

ALTER CONNECTION <connection_name> PRIMARY

Behind the scenes, Snowflake updates the URL to point to the account that has been promoted to serve as the primary connection, and your client connections are redirected to this account within 30 to 60 seconds. 

Note that your existing accounts’ URLs will continue to work.

GET STARTED

Snowflake Client Redirect is available on all Snowflake commercial regions and requires the Business Critical edition. Join the public preview by filing a support ticket. For more information, you can review the documentation here.