Sync in Practice: From Kingdee Distributed Transfer-Out to Wangdiantong Transfer Order — Inventory Transfer Strategy Deep Dive
What This Strategy Solves
In a multi-organization, multi-warehouse retail business, a common pain point is that "the ledger lives in Kingdee while the physical stock lives in Wangdiantong." Kingdee Cosmic handles financial books and the distributed transfer-out instruction, while Wangdiantong handles the actual inbound and outbound at stores and warehouses. When an intra-organization transfer happens, the source side generates a distributed transfer-out document in Kingdee, and a matching transfer order must appear on the Wangdiantong side almost immediately — otherwise the goods have already left the warehouse but the books cannot trace where they went.
The core value of this strategy is to push the transfer-out instruction downstream within milliseconds to minutes. We use the Qeasy Data Integration Platform to host this link. The key point is not just "it can be transmitted" but "it is transmitted correctly, stably, and traceably."
Data Flow and Field Mapping
The overall direction is B -> A: Kingdee Cosmic (B) -> intermediate layer -> Wangdiantong (A). The intermediate layer in Qeasy is responsible for field cleansing, code translation, and status tagging.
Key field mapping (excerpt — actual values depend on the customer's master data):
| Business Meaning | Kingdee Cosmic (Source) | Intermediate Layer | Wangdiantong (Target) |
|---|---|---|---|
| Document Number | FBillNo | Keep original, add prefix to avoid duplicates | External Order No. |
| Source Organization | FStockOrgId | Org code mapping | Warehouse Code |
| Destination Organization | FDestStockOrgId | Org code mapping | Target Warehouse Code |
| Material Code | FMaterialId | SKU master data match | Product Code |
| Quantity | FQty | Unit conversion (if needed) | Transfer Quantity |
| Business Date | FBusinessDate | Normalize to yyyy-MM-dd | Document Date |
| Document Status | FDocumentStatus | Push only approved | / |
Note: Do not scatter code mappings across strategies. We recommend building an "Org-to-Warehouse" master mapping table in Qeasy that all transfer-related strategies can share. Change it once and the change takes effect everywhere — this is one of the most common patterns used by Qeasy customers in the field.
How to Configure in Qeasy
- Data source registration: Connect Kingdee Cosmic as the source system, select "Distributed Transfer-Out Document" as the business object, and set the filter condition to "Document Status = Approved AND Modified Time >= Last Sync Time."
- Data target registration: Register the transfer order interface on the Wangdiantong side, use the external order number as the primary key, and enable idempotent writes to prevent duplicate documents on reruns.
- Mapping orchestration: Process the header and body in two stages. Land the header first, then append the body by FEntryId in batches. A single body-row failure should not roll back the entire document — this is the common "header-then-body staged" approach used by Qeasy customers.
- Centralized code mapping: Put org, warehouse, and SKU mappings into a single mapping center. Strategies only reference them, never hard-code values.
- Exception handling: Configure three branches — network jitter retries, missing-field alerts, and target-system validation failures routed to a manual queue.
Implementation Steps
We recommend a three-phase rollout: "Full Load first, then Incremental, then Steady State."
- Incremental starting point: First confirm the timestamp of "Approved" documents on the Kingdee side. The incremental cursor in Qeasy starts from this point to avoid pulling historical dirty data along with the new data.
- Full load trigger: Before going live, do a cold-start full load to backfill all in-flight transfer orders at once. Qeasy's full-load trigger runs on dual tracks — syncing while recording differences — followed by a manual reconciliation once the run completes.
- Scheduling frequency: Intra-organization transfers are time-sensitive, so we usually configure a 5-minute polling cycle plus an event-triggered channel. The event trigger relies on the Kingdee approval callback, while polling acts as a fallback. This is the typical "incremental and full-load dual-track" setup.
- Steady-state monitoring: Check the difference table daily for the first 7 days after going live, then switch to weekly inspections.
Lessons from the Field
- Misusing organization codes: Kingdee has separate fields for "Source Organization" and "Destination Organization," while Wangdiantong uses the "Warehouse" dimension. If only one side is mapped, the receiving side will never get the goods. This is a common pitfall — the safe approach is to validate both directions.
- Unit conversion hiding in the body: The same material may use different units of measure in different organizations. Mapping only at the header level misses this. FQty in the body must always go through a unit conversion function.
- Mixing approval status with posting status: Kingdee has both "Approved" and "Posted" statuses — only "Approved" should be pushed. A typical mistake is using posting as the filter condition, which causes duplicates because every upstream posting triggers a push.
- Overly broad idempotency keys: Using the document number as the idempotency key is fine, but if body rows also use the document number for deduplication, a partial row failure will prevent the entire document from rolling back. It is safer to deduplicate body rows using FEntryId alone.
- Time zone and date type issues: Kingdee date fields include a time stamp, while Wangdiantong only accepts the date. Pushing directly will be rejected by the target system, so be sure to truncate the value at the intermediate layer.
When This Applies and When It Does Not
Applies: Intra-organization transfers within the same legal entity, store-to-store transfers, routine replenishment between e-commerce warehouses and store warehouses. Does not apply: Transfers across different legal entities (which must go through a sales-outbound flow), transfers that require approval workflows (this strategy only pushes approved documents), and projects where master data has not yet been aligned (forcing synchronization before codes are standardized will amplify dirty data).