Integration Overview: Jiandaoyun and Kingdee Cloud Cosmic Sales Order Sync (16 Strategies, Phased Closure)

Scenario and Value

In one of our retail/manufacturing engagements, the business side kept running into the same wall: sales orders are entered in Kingdee Cloud Cosmic, but the approval step lives in Jiandaoyun. Inventory and delivery notices have to be chased manually in the ERP. When the post-promotion reconciliation finally happens, finance and operations never quite agree — not because anyone typed it wrong, but because there is no automatic closed loop between the two systems.

What made it worse: the customer runs multiple organizations (103/109/114). Each organization has different approval templates for sales orders, sample orders, and miscellaneous outbound orders. If the integration strategy is one-size-fits-all, cross-wiring is inevitable.

This integration addresses three concrete problems:

• Stable delivery of master data (materials, suppliers, customers, employees) from Kingdee to Jiandaoyun, as a prerequisite for business documents.
• Routing of sales orders, delivery notices, and outbound orders by organization and status, pushing only what is needed to the right Jiandaoyun form or approval node.
• Resilience against missing codes, duplicate documents, and network hiccups — observable, retryable, and recoverable, without polluting downstream systems.

We run this on the Qeasy iPaaS platform (轻易云数据集成平台), breaking the 16 strategies into three phases — master data, material update, and business documents — and orchestrating them into a single observable pipeline.

Integration Architecture and Data Flow

Kingdee Cloud Cosmic acts as the master data and transaction source, Jiandaoyun carries the business-facing forms, and the integration platform sits in the middle handling transport, transformation, scheduling, and monitoring.

Kingdee Cloud Cosmic (ERP)
  ├── Master Data: Materials / Suppliers / Customers / Employees
  └── Business Docs: Sales Orders / Delivery Notices / Outbound Orders / Misc Outbound
            │
            ▼
   ┌────────────────────────┐
   │  Qeasy iPaaS Platform  │
   │  · Schedule Orchestration
   │  · Code Mapping (FNumber etc.)
   │  · Incremental / Full Filters
   │  · Retry / Dead-Letter / Alert
   └────────────────────────┘
            │
            ▼
Jiandaoyun
  ├── Forms: Material / Supplier / Customer / Employee
  └── Forms: Sales Order / Approval / Misc Outbound

The data flow advances in three phases:

1. Phase 1: Master data readiness. Materials, suppliers, customers, and employees are pushed in parallel as the dictionary for subsequent transactions.
2. Phase 2: Material write-back. Using the mapping between Jiandaoyun's _id and King's FNumber, material changes are written back to Jiandaoyun to close the loop.
3. Phase 3: Business document routing by organization. Sales orders are split by organization (103/109) and status (approved/unapproved) into multiple parallel strategies; sample orders use an independent channel; delivery notices and outbound orders carry warehouse codes; misc outbound runs one strategy per organization (103/114).

Interface Inventory

The table below lists all 16 strategies. Field mapping must be tuned to the customer's actual dictionaries.

Implementation Highlights

Phased scheduling. Master data first; business documents parallel after master data is ready. A common pitfall is letting sales orders run before materials finish loading — downstream null pointers everywhere. The safe approach is to run a one-shot strategy that fully populates materials, then release business documents.

Incremental field selection. On the Kingdee side, use FCreateDate and FApproveDate as the incremental cursor; on the Jiandaoyun side, use updateTime. The FilterString should exclude voided documents to prevent sync-then-delete loops.

Full sync as fallback. Run a full sync on initial deployment or data repair; rely on incremental for daily operation. Master data is recommended for weekly full reconciliation as a sanity check.

Centralized code mapping. Material FNumber ↔ Jiandaoyun material code, customer FCustId ↔ customer code, supplier FSupplierID ↔ supplier code, warehouse FStockId ↔ warehouse, organization 103/109/114 must be centrally maintained. Business strategies should reference the mapping rather than hardcode values inside each strategy.

Tiered retry handling. Network timeouts and target-system 5xx errors use exponential backoff (30s/60s/120s, max 3 attempts); missing code mappings skip and log; business validation failures (e.g., duplicate document number) go straight to the dead-letter queue to avoid blocking the stream.

Privacy handling. All credentials between Kingdee and Jiandaoyun go through secret management; logs must mask sensitive fields such as phone numbers and customer names. The integration platform and target systems should connect over an internal network or private line; public-network scenarios must enable HTTPS and IP allowlists.

Best Practices and Lessons Learned

1. Centralize code mapping — never scatter it across strategies. Once FNumber is hardcoded inside a strategy, material retirement or code changes trigger cascading errors. On Qeasy, code mapping usually lives in a dedicated mapping table; business strategies reference it via lookup components, so one change propagates everywhere.

2. Split sales order strategies by organization × status. During one implementation, the customer reported "approved orders for organization 103 didn't sync." Root cause: a single big strategy used organization in (103,109) and status = approved, but a case-conversion in a Jiandaoyun field caused silent drops. Splitting into 4 independent strategies (Strategies 10–13) fixed the issue and made per-organization alerting much easier.

3. Two-phase write for header and line items. Sales orders have a header (customer, order number) and line items (material rows, quantities). One-shot commits have high rollback cost. On Qeasy, the typical pattern is to write the header first to obtain _id, then push line items using _id as the foreign key; either step can be resumed independently on failure.

4. Dual-channel routing for Qimen vs. non-Qimen warehouses. If the customer has both Qimen and non-Qimen warehouses for delivery notices, add a routing decision before Strategy 8 to dispatch to different mapping templates instead of mixing them together.

5. Don't guess monitoring thresholds. Pause dependent strategies after 3 consecutive master-data failures; alert on >10 missing-code records in a single run; alert on >10% per-strategy failure rate; manually process dead-letter queues with >100 entries; alert on sync lag >30 minutes — these thresholds have been validated across multiple projects and can be reused as-is.

When to Use Qeasy

If your scenario is "Kingdee Cloud Cosmic + Jiandaoyun + multi-organization + multi-status sales chain," with daily document volumes in the tens of thousands to hundreds of thousands, and you need stable incremental sync, strict code mapping, and an observable scheduling pipeline — the Qeasy iPaaS platform (轻易云数据集成平台) provides ready-made Kingdee and Jiandaoyun connectors, a built-in code mapping center, flexible schedule orchestration, and monitoring/alerting, materially reducing implementation and operational cost in multi-organization scenarios.