Guidewire Cloud Data Access Private Preview
Guidewire Cloud Data Access (CDA) is a streaming export service within the Guidewire Data Platform that captures changes from InsuranceSuite applications (PolicyCenter, BillingCenter, and ClaimCenter) and delivers processed records as Parquet files to an AWS S3 bucket. Cloud Data Access (CDA) also maintains a manifest.json file that records the last fully committed write timestamp for each table, indicating when a batch is complete and ready to read.
The Guidewire Cloud Data Access connector is designed specifically for CDA ingestion. It reads the manifest to discover tables and their committed timestamps, then ingests Parquet files only from timestamp folders that Guidewire has marked complete. This ensures that Fivetran syncs only consistent, fully written batches to your destination.
Key capability
Some of the key capabilities of Guidewire Cloud Data Access are:
| Capability | What it does | Why it matters |
|---|---|---|
| Manifest-gated ingestion | Reads the manifest before every sync and skips any timestamp folder Guidewire has not committed. Fivetran does not read the uncommitted folders. | Prevents ingestion of incomplete data. |
| Cursor-based incremental syncs | Tracks the last committed timestamp for each table and resumes from that point in subsequent syncs. | Keeps sync efficient and bounded. |
| Schema evolution | Automatically handles data type changes, new columns and tables based on your connection's schema change settings | Reduces manual maintenance. |
Features
| Feature Name | Supported | Notes |
|---|---|---|
| Capture deletes | ||
| History mode | ||
| Custom data | ||
| Data blocking | ||
| Column hashing | ||
| Re-sync | ||
| Row filtering | ||
| API configurable | API configuration | |
| Priority-first sync | ||
| Fivetran data models | ||
| Private networking | ||
| Authorization via API |
Supported deployment models
We support the SaaS and Hybrid deployment models for the connector.
You must have an Enterprise or Business Critical plan to use the Hybrid Deployment model.
Setup guide
Follow our step-by-step Guidewire Cloud Data Access setup guide to connect Guidewire CDA with your destination using Fivetran.
Sync overview
CDA file rows
Fivetran reads the committed Guidewire CDA Parquet files and loads the data into your destination without any modification. Fivetran preserves the Guidewire-specific columns such as gwcbi___operation in the destination exactly as CDA provided. Fivetran does not interpret these columns or convert them into inserts, updates, or deletes in the destination. The downstream transformation layer applies these changes.
Fivetran also adds file metadata columns, such as _file, _line, and _modified, and uses them as surrogate primary keys.
Manifest table
Fivetran maintains a MANIFEST table in your destination schema. During each sync, Fivetran upserts one row for each unique combination of CDA table and committed timestamp:
| Column | Data Type | Description |
|---|---|---|
table_name | String (PK) | CDA table name |
last_successful_write_timestamp | String (PK) | Committed timestamp from the CDA manifest |
total_processed_records_count | Long | Cumulative record count from manifest |
schema_history | JSON | Schema fingerprint history |
Sync notes
The following notes describe how the Guidewire connector processes Parquet files and handles re-sync operations:
- The connector only ingests Parquet files from committed timestamp folders. It always skips the folders that Guidewire has not committed.
- It supports connection-level and table-level re-syncs. During a re-sync, Fivetran reprocesses all committed CDA timestamp folders.
Configuration options
You can use a custom external_id parameter for authentication when creating a new Guidewire Cloud Data Access connection with the Fivetran REST API. We use the connection's group_id if you don't specify the value of the external_id parameter. Use the List All Groups endpoint to find the connection's group_id.
We don't support custom external_id values for connections created in the Fivetran dashboard. By default, we use the connection's group_id.