Skip to content
Cross-Department System Reuse: Why Changing the Warehouse ID Isn't Enough
R&D Management 19 min read

Cross-Department System Reuse: Why Changing the Warehouse ID Isn't Enough

This article details the critical checks needed when extending a replenishment system from one warehouse to another, covering product identity, local data sources, business rules, authorization, versioned releases, in-flight task transitions, rollback safety, and regression testing to avoid breaking existing deployments.

Data Bricklaying Diary
Data Bricklaying Diary
Data Bricklaying Diary

First Check What Can Be Reused, Don't Start by Copying the Entire Project

First compare product data from both sides. If they are indeed the same specification and item, reuse the confirmed SKU6204; do not create a new product identity just because the warehouse changed. Purchase orders, warehouse and supplier base definitions can also be shared after verifying their meanings.

However, Warehouse 1's inventory levels and which purchase orders count as its supply cannot be copied as Warehouse 2's data. Even when using the same system, queries must carry Warehouse 2's corresponding codes and check source, unit of measure, and access permissions.

Original definitions may also be insufficient. For example, Warehouse 1's documents only go to a single warehouse, while Warehouse 2 uses a document that allocates to multiple warehouses. You must clarify whether the whole document, line item, or allocation record corresponds to the warehouse, then decide how to extend the definition — you cannot just pick one warehouse to reuse the old field.

The first article discussed what can be shared and who maintains it. Now that Warehouse 1 is live, you must also check whether Warehouse 2's changes will affect it.

Having Warehouse 2 bring a few actual records for comparison reveals differences more effectively than simply asking "is your process the same?"

Product identity shared, inventory facts queried separately
Product identity shared, inventory facts queried separately

Same specification confirmed, share product identifier; each warehouse still queries by local code, source, and authorization; shared definitions do not copy inventory or procurement facts.

Same Products, Which Rules Differ

Verification Checklist

Product Identity : Whether same specification, whether code mapping is valid → Reuse same identifier for same object; model different products separately

Business Facts : Source of inventory and procurement quantities, whether units and timestamps satisfy purpose → Add local mapping and access conventions

Judgment Rules : Replenish by stock or procure by order, how period is determined → Select approved rules by scope

Create Purchase Order : Whether original interface supports Warehouse 2, whether creation status and attached operations are identical → Share same conventions where identical; separately confirm and test differences

Authorization & Approval : Local responsible person, quota, and executing entity → Confirm separately; cannot copy Warehouse 1 authorization

Experience & Examples : Whether experience applicability conditions still hold → Verify before use; supplement tests for different conditions

Ongoing Maintenance : Who manages shared definitions, who manages Warehouse 2 code mappings and rules → Assign specific owners; agree on who to notify and who confirms when changes occur

The same product can have different procurement purposes. Warehouse 2 may not procure all items by order, so rules cannot be selected by warehouse number alone; you must check whether this demand is for stock replenishment or for a specific order.

After confirming warehouse, purpose, period, and effective version, the program selects the approved rules; if unclear in discussion, ask first. If Warehouse 2 rules are not configured, or two rule sets match simultaneously, or no convention exists for which to pick, pause the related calculation and find the responsible person. Do not default to Warehouse 1's ready-made rules.

Reasonable differences can be retained, but factual errors must be fixed. For example, the same purchase line shows the same packaging basis on both sides, yet one converts a box to 10 pieces and the other to 12 pieces — you must investigate the conversion, not explain it away as "departmental caliber difference."

Same bearing, may be procured for different purposes
Same bearing, may be procured for different purposes

Select approved rules by this demand's warehouse, purpose, period, and version; pause and confirm when missing or conflicting; do not just swap warehouse numbers and apply existing rules.

Distinguish Updating Facts from Changing Definitions

A purchase order receives partial goods; the unreceived quantity decreases accordingly — this is a business record change. Update the corresponding quantities per the original system records; usually you do not redraw the "product" concept.

Which purchase order statuses count as supply, whether they must arrive in the current period — these are rules. To change rules, first confirm who they apply to and when they take effect, then verify that queries and calculations follow the change.

Supplier packaging changes require examining the document basis. An old order had 1 box = 10 pieces; new packaging is 1 box = 12 pieces. You cannot change a global conversion factor and let historical procurement quantities change with it. How to identify old vs. new packaging and when each applies must be traceable to the corresponding records.

Projects using OWL (Web Ontology Language) can use ontology IRIs and version IRIs to distinguish an ontology and its specific version; the IRI serves as an identifier; the import mechanism allows one ontology to reference another's definitions. [Ref 1] These mechanisms facilitate organizing shared content, but whether they can continue to be used in the original application still requires actual verification.

For example, importing a new version of definitions may succeed, but the old program still assumes each document has only one warehouse, leading to misuse at runtime. OPM (Object Process Methodology) models are the same: you must align confirmed models, code mapping tables, calculation programs, and application versions; you cannot just label everything "v2" and consider the release complete.

New packaging 12 pieces, old order 10 pieces cannot follow change
New packaging 12 pieces, old order 10 pieces cannot follow change

Here old and new documents each have their own packaging basis; conversions are verified separately; contradictions in the same record must still be corrected; new packaging cannot rewrite the historical meaning of old orders.

Before Release, Identify All Consumers

Suppose the shared unfilled purchase quantity filtering rule needs revision. First check what currently uses it: Warehouse 1 assistant, Warehouse 2 dashboard, scheduled replenishment tasks may all call the same calculation service; caches and tool documentation may also retain old results and explanations.

List these consumers and their owners, then confirm item by item what to change and when to switch. A single-warehouse change only takes effect in the agreed scope; but if a shared definition or program changes, you must check whether other consumers are affected — you cannot test only Warehouse 2 just because the request came from there. Historical queries may still need old rules; today's new business should follow newly effective requirements.

Also identify in-flight tasks. For example, a 60-piece suggestion is already approved, but the new rule requires recalculation. You cannot simply replace the quantity and resubmit after release. First decide per the business-confirmed transition plan whether to recalculate; if the result changes, re-verify the approval scope; if re-approval is needed, stop there. If the original creation request has already been sent, continue to follow the original request for results — do not open a new order just because the version changed.

Allowing old tasks to finish under old rules does not mean permissions and approvals are frozen together. Pre-submission checks of current conditions are still required. Each task retains the rule and program version it used; during transition, if a matching version cannot be found, pause the corresponding step — do not temporarily cobble together an untested combination.

Release Checklist

Change Reason : Which definition or fact handling was insufficient, and who confirmed the basis

Impact Scope : Warehouses, products, callers, caches, and in-flight tasks

Version Combination : Ontology and its referenced versions, OPM, code mappings, calculation and action programs, and corresponding tests

Switch Arrangement : Which scopes switch when, whether old parameters and outputs still apply, what to do with approved and submitted tasks

Validation & Approval : Actual results for old and new samples, uncovered items, and release owner

Rollback Conditions : Whether technical rollback still complies with current business rules, how to handle already executed actions

Rollback must be tested in advance. Before reverting to old code, verify that the old program can still read data, tasks, and states left by the new version. Code starting up does not mean tasks can continue.

Moreover, reverting the service to the old version will not make newly created purchase orders disappear. Documents needing modification or cancellation must follow the original system process. When new rules are already effective and old rules no longer apply, you cannot continue using old rules for calculation just to restore service quickly; you should pause affected capabilities and retain original requests for verification.

New version released, how to hand over old tasks?
New version released, how to hand over old tasks?

Confirm transition per task stage; do not force uniform version switch; pre-submission still checks current permissions and approvals. Before rollback, verify data, task compatibility, and current rules; code rollback does not cancel purchase orders.

Second Warehouse Must Run Its Own Test Cases

First lock down Warehouse 1's original 60-piece sample set. Parts where applicable rules have not changed should continue to produce the originally agreed results. Where rules actually changed, business personnel must separately confirm new expectations — you cannot both demand new rules and demand that numbers never change.

Then test with Warehouse 2's order-based procurement records, calculating expectations per its own rules; do not automatically fill to Warehouse 1's target quantity. Different packaging, missing corresponding rules, a single request matching two rule sets, and Warehouse 1 users querying Warehouse 2 data — all must be tested. Compliant Warehouse 2 requests should process normally; unauthorized or insufficient-basis requests are intercepted per convention.

On switch day, at least run through pending-approval, approved-but-not-submitted, and sent-awaiting-verification tasks separately. Observe which rule version they used, who needs re-approval, and whether the original request can still be traced. Historical records must be restorable per their original basis, not silently altered by the new version.

Even if the rule interface returns the same field names, meaning changes must be checked at callers. For example, an original dashboard interface's "unfilled purchase quantity" counted all valid documents; after integrating replenishment, it changed to count only portions arriving in the current period. Can the old dashboard title and explanation still be used? The program does not error, so such issues go undetected.

Record how much new integration, confirmation, testing, and maintenance work Warehouse 2 actually added, then discuss what reuse saved. Definitions were reused, but data and authorization may be more complex; the second time is not necessarily cheaper. Currently this is only deduction, so record discovered differences first — do not write projected savings as achieved results.

Warehouse 1 regression, Warehouse 2 validates per local rules
Warehouse 1 regression, Warehouse 2 validates per local rules

Warehouse 1's original functions must be regression tested; Warehouse 2 must also validate with its own samples. When rules change, re-confirm expected results; the diagram lists the planned checks.

Summary

When Warehouse 2 integrates, shared product definitions and existing programs need not be rebuilt. But first verify with local records: where data comes from, what rules govern this demand, who has approval authority.

After verifying differences, arrange version switch and old task handover. Especially for shared programs: Warehouse 2 fixed, you must still test Warehouse 1; otherwise the new warehouse works but the original warehouse breaks.

If numbers mismatch during operation, should you check model, data, or interface first? The next article will troubleshoot using an erroneous replenishment suggestion.

References

W3C, OWL 2 Structural Specification and Functional-Style Syntax, sections 3.1, 3.3, 3.4. Used to illustrate ontology and version identification and import mechanisms; compatibility, release, and rollback checklists are author-designed; version identification does not automatically provide these guarantees. https://www.w3.org/TR/owl2-syntax/

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

data modelingversion managementsystem integrationrelease processBusiness Rulescross-department reuserollback testing
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.