Pipelines increasingly feed systems that act on data without a person checking it first. That puts more weight on the controls around them.
CData Sync 26.4 adds those controls: vaults for secrets, schedules that follow demand, and read-only access for AI assistants. It also extends change data capture (CDC) support for SQL Server, DB2 for i, and SAP Gateway, and adds a direct-insert path for initial loads. Several of these changes started as requests from customers in healthcare, insurance, transportation, and automotive. This post covers everything released in Sync since 26.3, including our weekly releases.
Control that fits how you operate
Vault support
A data pipeline holds the keys to your most sensitive systems. Security reviews tend to come down to how many places those keys sit and who can read them.
CData Sync 26.4 adds vault support built on one rule: jobs reference secrets by name. It covers Job Event scripts and connection properties. Secrets can live in an encrypted Local Vault or come from host environment variables, and Local Vault is included in every edition.
Schedules that follow demand
A transportation company replicating DB2 for i into Snowflake wanted CDC to run more often on weekdays and less often on weekends to manage compute cost. A large healthcare services organization replicating Dynamics 365 into SQL Server runs a periodic full ReSync, because Dynamics 365 doesn't report deletes.
In Sync 26.4, a job can carry multiple schedule rules and a recurring ReSync. Cadence follows demand, and the ReSync runs on the job's own schedule.
Read-only access for AI assistants
An AI assistant is only as safe as the access you give it. CData Sync 26.4 includes a Model Context Protocol (MCP) server that lets an AI assistant look up workspaces, connections, jobs, transformations, and pipelines, pull their logs, and run a health check.
This first version is read-only, so it can look but can't change anything. It also runs under the signed-in user's identity, so an AI assistant sees only what that user's permissions allow. Secret values in connection details and scripts are masked.
CDC for SQL Server, DB2 for i, and SAP Gateway
Append-only destinations like Parquet can't merge, so ordering within a transaction has to be settled before data lands. When a reload procedure deletes a row and re-inserts it in one transaction, SQL Server records both changes under the same log sequence number (LSN).
CData Sync 26.4 settles it at the source. Within a single transaction, Sync collapses changes to the same key into one row, so a delete followed by an insert arrives as an update.
Change log: DELETE 42, INSERT 42, same LSN
Sync output: one UPDATE for key 42, with the new values
The same approach runs through the rest of the CDC work. Sync reads which columns SQL Server tracks and aligns the replication schema to them. It advances its checkpoint on every run, including runs that return no rows. It also captures DB2 for i one member at a time and adds SAP Gateway as a CDC source.
Direct inserts for initial loads
An initial load into an empty table is all inserts. For eligible tables, CData Sync 26.4 adds a direct-insert path that skips the merge when it reads in parallel into a target that is new, empty, or truncated. MySQL and Apache Doris destinations also gain a bulk-load path; the method MySQL's documentation recommends for loading large volumes.
Before you upgrade
SQL Server CDC emits one row per key per transaction. Intermediate changes within a transaction aren't sent, and a delete followed by an insert arrives as an update.
DB2 for i captures only the default member when you select a physical file. To capture another member, select a SQL alias for it.
Upgrade to CData Sync 26.4
CData Sync 26.4 is available now. For the full list of changes, including API endpoints, options, and settings, see the CData Sync documentation portal.
Existing customers: download the latest release and generate an updated license key through the CData Portal.
Evaluating Sync for the first time: start a free trial.
