How does the asset management software system fit in? Architectural stratification and integration points resolution
RFID Knowledge

How does the asset management software system fit in? Architectural stratification and integration points resolution

PublishedShanghai Zejie Smart Technology Co., Ltd.

A company that rents equipment, which at the end of the year found 200 devices on its books, found only 170 on site. The rest went to nowhere, and nobody said anything. The spreadsheets, each department, one account, purchase, bank, access, transfer to scrap, each link has data, but it's not right. It took six months to re-establish asset management software, turning chaos into a traceable process. Here we go by unturning the key points around the build.

First Cam: a first inventory of the system's boundaries

After that count, the project team sat down and reset it, finding out that the real problem was not "no equipment found", but that three layers were missing at the same time:

These three layers are the first of a system structure that later took shape – data to be centralized, process to be solid and interfaced. The construction of any asset management system is essentially addressing these three things, with the difference between doing it in depth or in a way that is not very simple.

Structure layer: break down the system to a few layers, and one thing at a time.

The purpose of the architecture is not to "seem professional" but to make changes at each level do not affect other layers. For example, an interface should change style, not move to a database structure; processes should be adjusted, not to a hardware interface. How exactly does that work, the more common approach in the industry is four levels:

Performance layer: human interface

The principle is that the interface is as thin as possible, all "judgements" are done in the business level, and avoids a multiple error rule, and then it's left out.

Business logic level: System brain.

The additions and deletions, depreciation calculations, state flow, approval rules are all at this level. It is the process implementer that determines how a piece of equipment can be moved from the purchase to the final end of life, how it should go, who has the right to approve and under what conditions.

Asset management and inventory management seem close, the difference is that assets have a "life cycle" -- depreciation, transfer of responsibility, value assessment. Business levels can be written in terms of inventory, post-depreciation, life cycle supplement, change costs are high. It is recommended to model at the beginning of the asset's life cycle.

Data access level: middleman connected to database

This layer deals with databases: query, write, service control. The advantage of independently taking is that if in the future a database is to be replaced by a relationship database, or read-written separate (segregated Query and write-in for performance), then change is enough at this level, without any involvement in business logic.

Data Layer: Main data should be standard

The source of the "three numbers for the same device" in the previous case is at the data level. Primary data (the most basic, unmistakable data in the system, such as asset numbering, asset name, classification, unit of measure) are prerequisites for the whole system to be able to speak with each other.

The lesson is: Master data do not seek to be in place once, but the "Main Data Management" module must be maintained. Even if the first version only contains asset numbers, names, categories of three fields, it also ensures that these three fields have a single source -- they cannot be added at any time at business level, and they need to go through the main data process.

Layer Duties Common pedals.
Performance Level Interface presentation and operation capture Write the validation logic in the interface, and the re-end will be disabled.
Business logic layer Processes, rules, state flow Modelled along the lines of inventory, with post-retrogressive depreciation unchanged
Data Access Layer Database reading and writing, transaction control SQL is in the business code.
Data Layer Main data, basic data Multiple sets of numbers for the same asset, not matched.

Integrated elements: how to relate to existing systems

Asset management is never an island. A new system that cannot be financially, HR, ERP up, value of assets, accountability, inventory connection is completely broken. Integration is the key to whether it's going to land, and where it's most problematic.

Interface before interface, data before process

And so many projects look at the interface as soon as it's done, and then they find that the financial depreciation rules are not available, and HR's duty bearers are not synchronized. The correct sequence is: how do you get key data down, then draw flow charts, and then make interfaces.

Take an example of a manufacturing client whose assets are purchased from ERP purchase orders, and the asset is automatically brought out of the supplier's inventory, the purchase price, the contract number. If the ERP doesn't work, each warehouse must be refilled, and no one will use the system after three months.

A few types of systems that are commonly integrated.

How do you choose the three ways of integration?

Real-time interfaces, batch synchronization, mesosheets, with applicable scenarios in each of the three approaches:

It is recommended that the project run through the middle watch at the beginning, and then gradually reach the real-time interface after it has stabilized. Shanghai & China Tech observed in its past projects that the real-time interface was done from the moment of arrival, and debugging cycles tend to be 2-3 times longer than the intermediate-table programme.

After the line: Runs in a bad way. Look at these details.

The system is not the end, but the beginning of a real test. The structure is so beautiful that you can't run or run or you can't run. A few details must be kept close to the line:

The main data governance is long-term engineering

The main data is not automatically clean because of the system. The rental company mentioned above, has a dedicated data governance team on line, which checks new asset numbers, renames assets, and classifies unusual records once a week. It's three months before the master data quality improves significantly.

The inventory cannot be based on a system alone.

The system can tell you how much money is on the books, but it cannot tell where the physical items are. The core of the inventory is "the reconciliation ratio", which recommends a small-scale disk per month, a full-size, quarterly basis, and an analysis process for the discrepancy count, which cannot be simply a numerical change.

The power is not to be relaxed.

Asset data are about the amount and person responsible, and the authority must be fine. It is recommended that it be controlled on three levels: "Performance-Function-Data Range" -- sensitive operations (deleted, value adjusted, end of life) with a second approval or log mark. One audit found that a group of assets had been substantially modified by the person responsible, but then checked for the logs, because the rights were designed too broadly.

Report and interface preset extension

Today's asset desk account, which may be depreciated tomorrow, idle ratio analysis, lease-assets return statement. The report engine suggests a configurable way to not encode each report (by writing statistical logic in the program). Interface fields also have to set an extension, and avoid adding one field to another.

Back to the point of departure: build-up is not a technical issue, but a management issue

Back to the first count, and finally they actually do it, not how beautiful a structure is, but three things are solid: master data is managed, process is followed, systems and financial HR can talk. Software is just a tool, without a corresponding management action, and a good structure cannot afford a proper inventory.

The process of building an asset management system is essentially a digital combo of the management process. If you're thinking about doing this, you can ask three questions: What is the most painful part of the story? Which data must be broken? Can the current process run through paper first? Think clearly of these three, and then talk about technology, and there will be less curves.