Why Data Cloud’s Zero Copy Makes Business and Technical Sense

Enterprises now store more data than ever. But data gaps still exist between data warehouses and business applications, including for Sales, Service, Marketing and Commerce. This challenge grows exponentially as AI agents are used in business applications. Customer interactions, automation and AI-driven insights require real-time access to data, but traditional integration methods including ETL or APIs create latency, complexity, and friction. Salesforce Data Cloud bridges this gap with Salesforce zero copy.

Zero copy is a data federation technology that enables enterprises to access and query data without copying it. While some form of data federation has existed for years, adoption has been slow due to complexity and lack of seamless integration into enterprise platforms. Salesforce Data Cloud’s zero copy, however, differentiates itself not only because it removes the need for traditional ETL data replication, but also because it can scale to meet diverse business application and agentic needs. This means organizations can power all their applications with the freshest data—without costly, time-consuming data movement.

Data Cloud zero copy solves a number of key problems that enterprises face, including:

• Harmonized data across sources via metadata

Traditional approaches often create data silos, making it difficult to unify and analyze data across Sales, Service, Commerce and Marketing platforms. zero copy eliminates this by enabling direct access to data without duplication or migration, keeping it within its original secure environment, and alleviating the risks and complexities of managing and migrating personally identifiable information (PII).

• Real-time insights and data synchronization

With zero copy, businesses can query data across multiple systems in real time, ensuring that decision-making is always based on the most up-to-date information.

• Simplified monitoring and maintenance

Legacy approaches require constant monitoring to address failures in data copying processes. zero copy removes this burden as data is not copied over.

• Agentic AI apps powered with all your data and governance parameters

Agentforce agents for customer experiences, sales, service, marketing, commerce, AI, analytics, and automation can access your enterprise structured and unstructured data via zero copy, with Governance, to power AI-based experiences internally and externally.

Let’s face it. Depending on data volume and velocity, sometimes a physical copy is the best answer and other times leaving data where it lives. Salesforce zero copy complements existing Data Cloud capabilities to ingest data into the Data Cloud Data Lake.

Data Cloud customers have the choices below about how to access data and optimize performance and cost.

• Physical ingest: Depending on the type, velocity, and volume of the data you can always choose to physically ingest a copy into Data Cloud.
• Zero copy Live Query: In the true nature of zero copy, Live Query allows Data Cloud to access your Data Lake’s tables dynamically, pulling only the required data based on system design and specified criteria. This streamlined method supports critical operations such as segmentation, unification, and the creation of actionable insights. This data is not hosted in Data Cloud’s Data Lake and the only records that persist are the Data Cloud curated views, such as Unified Individual.
• Zero copy Cached Acceleration: While zero copy does remove the need to persist External Data Lake tables in Data Cloud, that doesn’t mean that there aren’t processes and scenarios where temporarily persisting or caching a copy of the data may still be necessary or advantageous. With Cached Acceleration, Data Cloud can temporarily cache the external data and periodically and incrementally refresh that copy. You can decide what option is best on a per-stream basis, avoiding a one-size-fits-all approach for each table or source. This is an optional optimization, not a requirement. When data caching is recommended, it’s to enhance query performance for large datasets or to help clients control cost based on query volume.

Data Cloud’s zero copy is bi-directional, meaning the data from external data lakes is available as if it were natively stored within Data Cloud. Additionally, the enriched, unified data from Salesforce Data Cloud can be effortlessly integrated or shared back into your data lake without copies, eliminating the need for any outbound ETL processes.

1. Data in (data federation): External data sources (e.g., data warehouses or lakes) are queried live by Salesforce Data Cloud, which lets you access and use data without copying it into Salesforce systems. This federation capability works in two primary modes: query federation and file federation.
2. Data out (data sharing): Insights generated within Salesforce Data Cloud (e.g., segmentation, identity resolution, analytics) can be accessed by external platforms like Snowflake in real time, maintaining synchronization without duplication.

Using advanced metadata management and query pushdown techniques, zero copy allows you to leverage tables from all major hyperscalers such as Snowflake, Data Bricks, Redshift or Big Query in Data Cloud without having to create, persist and host a copy of the selected tables in Data Cloud. Zero copy supports not only querying external data in Data Cloud—it allows external data lakes to query enriched or unified Salesforce Data Cloud tables, and generate insights in real time, without outbound duplication. This dual capability is distinctive to Data Cloud.

Salesforce Data Cloud leverages Apache Iceberg, a high-performance open table format, to store, manage, and process large-scale data efficiently. Iceberg is designed to handle petabyte-scale datasets across cloud storage while enabling fast queries, optimized data versioning, and seamless data lakehouse architecture integration. Iceberg enables Salesforce zero copy, which allows external data to be federated into Data Cloud without moving or duplicating it. This means Data Cloud can query data where it resides, leveraging Iceberg’s indexing and metadata management for optimal performance.

Additionally, Iceberg plays a crucial role in File Federation (In Beta Now), allowing Data Cloud to directly access external storage without involving the external system’s compute resources, making it a foundational technology for zero copy’s ability to query and integrate data at scale.

Zero copy offers two approaches for federated data access.

1. Query federation

Query federation enables Data Cloud to access external data by federating queries directly to the external source via JDBC drivers. For example, when a query in Data Cloud requires data, the query engine identifies the external source and leverages the Hyperscaler’s JDBC driver to fetch the data dynamically. The data is processed in memory and discarded after use—no data is stored in Data Cloud.

2. File federation (In Beta Now)

File federation operates similarly to query federation but does not involve the external source’s compute resources. Unlike query federation—where the external system processes the query—file federation allows Data Cloud to interact directly with external storage without engaging the compute layer of the external data warehouse. This approach provides efficient access to large datasets while preserving performance and cost efficiency.

Salesforce CRM data is synched in near real-time to Data Cloud. Any updates—whether it's a new customer record, an opportunity change, or a support case—are reflected instantly in Data Cloud. This near real-time synchronization ensures that your most critical business data is always up to date and can be effortlessly joined with zero copy data from your external warehouses. All Salesforce applications (including Sales Cloud, Service Cloud, and Commerce Cloud) can leverage Zero Copied data for AI-powered insights, real-time personalization, llow automation, and agent recommendations. Think of this as double zero copy–once data is accessed in Data Cloud via zero copy, it can be leveraged across the Salesforce ecosystem without being physically stored again. This deep integration ensures real-time insights and consistent data availability nda maintains governance and compliance standards.

Conversely, since zero copy is bi-directional, you can also project enriched data, AI-driven insights, or CRM-warehouse-joined datasets back to your data warehouse. With Data Cloud One, you can reference and use federated data across multiple connected Salesforce orgs.

Data Cloud zero copy lets you access data where it lives without a copy but what happens when a table in an external system accessed via zero copy is changed—for example, a column is renamed or maybe deleted? This is where Data Cloud shines. Data Cloud explicitly separates the physical data model from the logical, business-friendly data model using Data Model Objects (DMOs). DMOs isolate downstream data consumers such as identity, segmentation, data activation and analytics from changes in the physical layer known as Data Lake Objects (DLOs).

Collectively, these Data Cloud Data Model Objects and the built-in, AI-powered Data Cloud data mapping experience are known as Data Cloud Harmonization. Harmonization enables customers to build a business-friendly logical data model that normalizes data nomenclature and isolates critical business systems from physical data.

Without this harmonization, a change in an external Snowflake table could break hundreds of dashboards or segments. This is a key weakness of ‘composable’ solutions that expose physical data models to critical business systems, resulting in unexpected maintenance costs and outages. Data Cloud zero copy works seamlessly with Data Cloud Harmonization and protects you from these real-world situations where physical data is changed by one team without any awareness of the downstream impact.

With zero copy, data governance between Salesforce applications and external data sources can be managed within Data Cloud. This is critical because Salesforce application users, such as sales and service reps, do not interact directly with Snowflake or other data lakes.

By mapping warehouse metadata into Data Cloud, zero copy enables governance policies to be centrally defined and enforced in Data Cloud, ensuring secure and seamless access across Salesforce applications and external data sources.

Data Cloud zero copy delivers on performance and security using a number of strategies including Advanced Query Pushdown for Query Federation, File Federation and Secure Data Transfer.

Here’s how:

• Advanced query pushdown: Instead of creating full data replicas hosted in Data Cloud’s Data Lake, zero copy delegates data retrieval to the originating data warehouse or lake using an optimized pushdown query that retrieves just the data needed. This query pushdown optimization is critical and is what differentiates Salesforce zero copy from other solutions. This ensures streamlined data transfer leveraging the performance of platforms like Snowflake for query execution while minimizing data retrieval and response times.
• Secure data transfer: Zero copy handles the Data Lake to Data Cloud security end-to-end, relieving the integration team of this responsibility. Extra security can be enabled by leveraging Private Connect for Data Cloud, which enables Data Cloud to reach data sources locked down in an AWS VPC, such as Redshift, Snowflake or Databricks.

Salesforce Data Cloud was first to market with zero copy including out-of-the-box support for Snowflake, Databricks, Redshift and Big Query and we will continue to expand our support for key data lake vendors including Microsoft and others, but in order to maximize the zero copy ecosystem, Salesforce introduced the zero copy partner network in 2024—an ecosystem dedicated to secure, bidirectional, zero-copy integration with Salesforce Data Cloud. The initiative launched with initial partners Amazon Web Services (AWS), Databricks, Google Cloud, and Snowflake, and later welcomed Microsoft and IBM, all committed to enabling seamless zero-copy integrations. The partner network aims to continuously expand its partnerships while optimizing integrations to deliver exceptional security and usability.

Our Partner Network also includes 15 data ecosystem partners with seamless and secure integrations to their data lakes and ISV data kits through our zero copy network. These partners provide access to high-value external data sources that can enhance analytics, AI models, and business insights. The ecosystem partners include Dun & Bradstreet, Moody's, ZoomInfo, Workday, and The Weather Channel, which can be leveraged to enrich your Customer 360 for deeper intelligence and more comprehensive customer insights.

Salesforce Data Cloud’s zero copy is not merely a technical feature; it’s a strategic enabler for businesses aiming to achieve real-time insights and unified data management across their entire data estate. Zero copy addresses the critical challenges enterprises face today, offering scalability, flexibility, and unmatched integration within the Salesforce ecosystem and beyond. By offering the choice of how and when to ingest data, architects will be able to choose the best method to align to their business and operational needs. Salesforce Data Cloud continues to push the boundaries of what’s possible in data federation and data sharing, making it the superior choice for organizations seeking to harness the full potential of their data.

Salesforce Data Cloud’s zero copy is not only innovative but also a practical solution that tackles real-world data challenges, expedites time to value, and reduces integration costs.

Handle Massive Data Volumes: Zero copy’s ability to process billions of rows in real time has been validated in live customer environments, processing over 11 trillion records from external sources in the past 11 months. Any suggestion that zero copy is incompatible with high-volume datasets is outdated.

Metadata Management Layer: Abstracts the origin of data for seamless integration, schema layer ensures that the marketer or end-users is shielded from underlying technical details or changes in data lake structure. Allows governance of Data that is accessed by users.

User-Friendly Setup: While initial implementation may involve effort, the long-term benefits of unified, real-time insights far outweigh the upfront investment.

Scalable Solutions for Enterprises: Salesforce Data Cloud caters to the needs of data-intensive enterprises, delivering unmatched scale and performance.