Salesforce debuts Zero Copy Partner Network to ease data integration

At Salesforce World Tour NYC today, Salesforce unveiled a new global ecosystem of technology and solution providers geared to help its customers leverage third-party data via secure, bidirectional zero-copy integrations with Salesforce Data Cloud.

“The big thing we hear from many of our customers is, ‘I already have data in N number of XYZ data platforms. How do I use that data within the context of all of these products that you’re offering, but in a way that I don’t have to create multiple iterations of that data and lose data lineage?’,” said Tyler Carlson, VP of business development and strategic partnerships at Salesforce.

Currently, a handful of startups offer “reverse” extract, transform, and load (ETL), in which they copy data from a customer’s data warehouse or data platform back into systems of engagement where business users do their work.

“The challenge that a lot of our customers have is that requires you to copy that data, store it in Salesforce; you have to create a place to store it; you have to create an object or field in which to store it; and then you have to maintain that pipeline of data synchronization and make sure that data is updated,” Carlson said.

The zero-copy pattern helps customers map the data from external platforms into the Salesforce metadata model, providing a virtual object definition for that object.

“It works in Salesforce just like any other native Salesforce data,” Carlson said. “You can do flows and automations on it. You can use it within Prompt Builder in our generative AI tools. You can have it just referenced on the record. It’s a much more seamless process for customers than having to purchase a third-party reverse ETL tool or manage some sort of pipeline back into Salesforce.”

Streamlining third-party data connections

Dubbed the Zero Copy Partner Network, the new ecosystem brings together ISVs and SIs in an effort to get rid of the custom integrations and complex data pipelines previously required to integrate and move data from Salesforce to external data warehouses and vice versa. Carlson said zero-copy integration offers businesses a more efficient, secure, and user-friendly way to connect data to business applications compared with traditional ETL processes and data pipelines.

The traditional methods of accessing data from third-party systems require businesses to build and maintain the connection and reconcile data changes over time. Zero-copy integration means teams access data where it lives, through queries or by virtually accessing the file.

Salesforce pioneered zero-copy bidirectional integrations with Data Cloud via partnerships with Amazon Redshift, Databricks, Google Cloud’s BigQuery, and Snowflake. With today’s announcement, integrations with BigQuery and Snowflake are generally available. Carlson noted that integrations with Redshift and Databricks are still in pilot but will go live later this year.

“We’re expanding the network to include our ISV ecosystem, enabling them to build on top of those zero-copy connectors to offer enrichment datasets, SaaS applications, and business applications with zero-copy integration to the data that resides in their platforms,” he added. “We are also extending that to our SI ecosystem to make sure that all of the global SIs we work with are certified and ready to help our customers with this distributed zero-copy integration pattern.”

According to Carlson, the benefits of zero-copy integrations include:

• Accessing live external data without copying (no reverse ETL). Zero-copy integration means accessing live data where it lives. If the source data changes, it updates everywhere immediately.
• Acting on data from anywhere in the flow of work. The data becomes part of Salesforce’s metadata framework and can thus be used in multiple ways, including generating BI or AI insights, marketing segmentation or activation, or creating unified customer experiences. For instance, a Data Cloud-triggered flow could update an account manager in Slack when shipments in an external data lake are marked as delayed.
• Sharing Customer 360 insights back without data replication. Not only does zero-copy integration provide the ability to read data from data warehouses and data lakes, it also enables sharing insights back to those systems without data replication.
• Maintain governance and security. Zero-copy integration eliminates the need for manual data movement, preserving data lineage and enabling centralized control fat the data source.
• Ground generative AI. Zero-copy integrations can connect to unstructured data (like PDFs, call transcripts, and emails) in addition to structured data, making it possible to bring unified business data into any AI prompt.

Currently, Data Cloud leverages live SQL queries to access data from external data platforms via zero copy. Carlson said that makes sense for data warehouses that have a query interface, but Salesforce also plans to use Apache Iceberg to provide a new, low-latency way of virtualizing data in Data Cloud with direct file access at the storage level. When released, this will extend zero-copy data access to any open data lake or lakehouse that stores data in Iceberg or can provide Iceberg metadata for its table.

Later this year, Salesforce will also add zero-copy support to the Data Kits that ISVs use to distribute datasets and enrich customers’ data in Data Cloud. Data Kits are containers of Data Cloud metadata that can be packaged and installed from the Salesforce AppExchange. The zero-copy support for Data Kits will enable customers to access third-party datasets from partners via the same zero-copy integrations they use for accessing their own data platforms.

“What it allows a third party to do is to do the heavy lifting of mapping that data into Salesforce’s canonical Customer 360 data model, so when a customer downloads that Data Kit, they get this rich data from a third party, but it’s pre-mapped so it’s zero ETL into the Data Cloud data model,” Carlson said. “We’re announcing the extension of that Data Kit framework to include the zero-copy connectors that we’ve built with our technology partners.”

For example, a partner like The Weather Company could offer a third-party Data Kit of real-time weather data with zero-copy support. An insurance company could procure that data set to support a gen AI application that generates email alerts for customers about an impending weather event.  With zero-copy support, the insurance company wouldn’t have to load that weather data into their platform. It could just reference the data virtually from The Weather Company’s data set.

Salesforce also announced zero-copy support for Heroku Postgres. Carlson claimed this will allow developers to leverage Heroku Postgres data across Data Cloud-powered Salesforce experiences and existing Heroku applications.

“We have heard from our customers about unlocking trapped enterprise data for use in CRM and grounding generative AI,” he said. “This zero-copy pattern we’re seeing is going to become more and more prevalent.”