Classification Processing Location

Processing Location controls where an ALTR Native classification job executes. Snowflake ALTR Native jobs choose between two locations, ALTR Hosted (default) and In-Warehouse; Databricks jobs always run ALTR Hosted; OLTP jobs always run on the OLTP Classification Agent. Which classifier rule conditions evaluate, and whether sampled data leaves your environment, depends on the location.

Note

Processing Location is a separate, job-level setting from sample size — see Classification Settings for how sampling works.

Condition Availability by Processing Location

Condition Type	ALTR Hosted	In-Warehouse	Classification Agent (OLTP)
Regex Conditions	Yes	Yes	Yes
Column Name Conditions	Yes	Yes	Yes
Data Location Conditions	Yes	Yes	Yes
Column Size Conditions	Yes	Yes	Yes
Data Length Conditions	Yes	Yes	Yes
Column Content Conditions	Yes	No	Yes
Snowflake Conditions	Snowflake only¹	Yes	No
Amazon Comprehend Conditions	Yes	No	Yes²
Google DLP Conditions	Yes	No	Yes²

¹ Databricks jobs, which always run ALTR Hosted, don’t evaluate Snowflake Conditions — only Snowflake ALTR Hosted jobs do. ² Requires your own Google Cloud or AWS credentials, configured on the OLTP Classification Agent.

A condition that can’t be evaluated for a given job is treated as not evaluated rather than failing the job — vacuously satisfied under AND logic, non-contributing under OR logic.

ALTR Hosted

With ALTR Hosted processing, ALTR samples your data source over the network and evaluates your classifiers in ALTR’s own infrastructure. This is the default processing location for Snowflake ALTR Native jobs, and the only processing location Databricks jobs use — Databricks has no In-Warehouse option; that processing model is specific to Snowflake.

Because sampling happens over the network, a scrambled sample of your data leaves your environment and transits ALTR’s infrastructure. When a classifier includes a Google DLP condition, that same scrambled sample also passes to Google’s Sensitive Data Protection service — see Google DLP Conditions for how that sample is constructed.

In-Warehouse

In-Warehouse processing runs an ALTR Native job directly inside your Snowflake warehouse. ALTR deploys a classification function and stored procedure that sample rows, evaluate your classifiers, and record results using your warehouse’s own compute — only the classification results return to ALTR for the report. Sampled values never leave Snowflake. Choose it as the Processing Location when you run a Snowflake ALTR Native job; it isn’t available for Databricks or OLTP.

Classification Agent (OLTP)

OLTP data sources always classify through the OLTP Classification Agent, a container that runs in your own environment. The agent samples data locally and evaluates your classifiers there — sampled data never leaves your environment, except for a classifier’s Google DLP or Amazon Comprehend condition, which calls the corresponding cloud service using credentials you provide rather than ALTR’s.