Breaking News
Databricks on Tuesday launched LTAP, a new data architecture designed to bring AI, operational and analytical data onto a single platform, eliminating the ETL pipelines and data copies that enterprises have long relied on to move information between applications.
The company said the new Lake Transactional/Analytical Processing (LTAP) architecture provides a common data foundation for AI applications and agents, allowing them to read, reason and act on enterprise data in near real time without the complexity of maintaining separate transactional and analytical systems.
The launch reflects a growing challenge for enterprises as AI adoption accelerates. Traditional data architectures were built around separate operational databases for business applications and analytical platforms for reporting, connected through change data capture (CDC) and ETL pipelines. While that model worked for conventional applications, Databricks argues it has become a bottleneck for AI systems that continuously consume and generate data.
"For decades, complicated data infrastructure was a tax that teams were forced to pay," said Ali Ghodsi, co-founder and CEO of Databricks. "Then agents arrived. In a matter of months, organizations effectively doubled their workforce, just not with humans. LTAP removes that bottleneck."
Rather than combining transactional and analytical workloads in a single engine or masking synchronization through "zero ETL" approaches, LTAP unifies data at the storage layer. Transactional applications continue to run on PostgreSQL while analytical engines access the same underlying data stored in open formats such as Delta and Iceberg.
According to Databricks, this allows operational data to become immediately available for analytics while enabling transactional and analytical workloads to scale independently without sacrificing performance or maintaining duplicate copies of data.
The announcement builds on Lakebase, Databricks' PostgreSQL-native transactional database, which the company introduced to bring operational workloads onto object storage. Databricks said Lakebase now supports thousands of customers, including Block, Ensemble, Superhuman and Zillow, and processes around 12 million database launches every day.
The company also unveiled new enterprise capabilities for Lakebase, including cross-cloud disaster recovery, Git-style branching and snapshots for safer database experimentation, and autonomous database operations that enable AI agents to monitor performance, identify slowdowns, recommend indexes and assist with recovery.
Databricks positions LTAP as a response to the rapid rise of agentic AI, where software agents increasingly perform tasks that require constant access to live enterprise data. As organizations deploy more AI-driven applications, maintaining multiple copies of operational and analytical data can increase costs, create synchronization issues and delay decision making.
The new architecture is built around three principles: a single governed source of truth managed through Unity Catalog, independent scaling of transactional and analytical workloads, and the elimination of ETL pipelines and replica infrastructure. The company says this approach simplifies governance while ensuring AI applications always operate on current data.
Industry observers see this as part of a broader shift toward AI-native data infrastructure, where enterprises seek to reduce complexity while enabling real-time AI workloads across business applications.
"For the health systems we serve, speed and accuracy in the revenue cycle directly affect their ability to deliver care," said Grant Veazey, CTO of Ensemble. "Lakebase and LTAP extend that foundation by unifying operational and analytical workloads on a single layer, giving our AI systems real-time access to the data they need."
As enterprises move beyond AI experimentation toward production deployments, vendors are increasingly redesigning data platforms to support AI agents that continuously interact with enterprise systems rather than simply analyze historical information. By eliminating the traditional divide between operational and analytical data, Databricks is betting that a unified architecture will become a foundational requirement for AI-driven applications.
LTAP will be available as part of Lakebase, with Databricks saying the new architecture is coming soon.
The company said the new Lake Transactional/Analytical Processing (LTAP) architecture provides a common data foundation for AI applications and agents, allowing them to read, reason and act on enterprise data in near real time without the complexity of maintaining separate transactional and analytical systems.
The launch reflects a growing challenge for enterprises as AI adoption accelerates. Traditional data architectures were built around separate operational databases for business applications and analytical platforms for reporting, connected through change data capture (CDC) and ETL pipelines. While that model worked for conventional applications, Databricks argues it has become a bottleneck for AI systems that continuously consume and generate data.
"For decades, complicated data infrastructure was a tax that teams were forced to pay," said Ali Ghodsi, co-founder and CEO of Databricks. "Then agents arrived. In a matter of months, organizations effectively doubled their workforce, just not with humans. LTAP removes that bottleneck."
Rather than combining transactional and analytical workloads in a single engine or masking synchronization through "zero ETL" approaches, LTAP unifies data at the storage layer. Transactional applications continue to run on PostgreSQL while analytical engines access the same underlying data stored in open formats such as Delta and Iceberg.
According to Databricks, this allows operational data to become immediately available for analytics while enabling transactional and analytical workloads to scale independently without sacrificing performance or maintaining duplicate copies of data.
The announcement builds on Lakebase, Databricks' PostgreSQL-native transactional database, which the company introduced to bring operational workloads onto object storage. Databricks said Lakebase now supports thousands of customers, including Block, Ensemble, Superhuman and Zillow, and processes around 12 million database launches every day.
The company also unveiled new enterprise capabilities for Lakebase, including cross-cloud disaster recovery, Git-style branching and snapshots for safer database experimentation, and autonomous database operations that enable AI agents to monitor performance, identify slowdowns, recommend indexes and assist with recovery.
Databricks positions LTAP as a response to the rapid rise of agentic AI, where software agents increasingly perform tasks that require constant access to live enterprise data. As organizations deploy more AI-driven applications, maintaining multiple copies of operational and analytical data can increase costs, create synchronization issues and delay decision making.
The new architecture is built around three principles: a single governed source of truth managed through Unity Catalog, independent scaling of transactional and analytical workloads, and the elimination of ETL pipelines and replica infrastructure. The company says this approach simplifies governance while ensuring AI applications always operate on current data.
Industry observers see this as part of a broader shift toward AI-native data infrastructure, where enterprises seek to reduce complexity while enabling real-time AI workloads across business applications.
"For the health systems we serve, speed and accuracy in the revenue cycle directly affect their ability to deliver care," said Grant Veazey, CTO of Ensemble. "Lakebase and LTAP extend that foundation by unifying operational and analytical workloads on a single layer, giving our AI systems real-time access to the data they need."
As enterprises move beyond AI experimentation toward production deployments, vendors are increasingly redesigning data platforms to support AI agents that continuously interact with enterprise systems rather than simply analyze historical information. By eliminating the traditional divide between operational and analytical data, Databricks is betting that a unified architecture will become a foundational requirement for AI-driven applications.
LTAP will be available as part of Lakebase, with Databricks saying the new architecture is coming soon.
See What’s Next in Tech With the Fast Forward Newsletter
Tweets From @varindiamag
Nothing to see here - yet
When they Tweet, their Tweets will show up here.
