On the development of ERP system: Data processing structure-traffic flow and data model

In the development process of ERP system, or use I prefer to say, in the production process of ERP, a lot of operation is the flow of data, data operation, data continue flow, this similar pipeline data processing,

For example, the processing of orders, to go through order fetching, order status check, order Financial Review, Order manual review, order logistics number acquisition, order delivery, and so on, is:

1-2-3-4-5-6-7-8,1 is our initial data, 8 is our final result, of course, in the actual operation process, not a single line so simple, a lot of lines are cross-mixed, so that the whole system looks more complex,

System software is the world's abstraction, we this time is simple, through the infinite cycle, as well as the small differences that occur in the cycle, long ago Our world, then, so to speak, our assembly line, we can also be divided, so that his complex appearance to simplify, this method is to find the intrinsic mechanism

One.

ERP Pipeline Model

1-->2-->3-->4-->5-->6-->7-->8-->9-->10, this is the simplest pipeline process, but in practice the source of the data may not be a line

But:

1------;} ====>3

2------;}

That is, 3 of the data source, two, 1 and 2, the same, 3 products can have multiple results, so the source of n data into the 3,3 operation data, produced n results, so that the pipeline becomes complex

But original aim, what we need to analyze is a single data processing model.

Two:

Data processing model

1---} **********{----->4

}====> * * ***3****}===>{

2---} **********}------->5

The above model, 3 for a data processing black box, 1 and 2 can pass data to 3, of course, 1 and 2 of the data type is different, for example, the processing of orders may be a consumer-to-consumer site, or an order from the Amazon, so the data needs to have data type, 3 after receiving data, data processing, According to different operation algorithm, product different results, corresponding to 4 and 5, for example, after the order audit, some orders are successful, some orders are unsuccessful, then the successful order, will go to the next step, but the failure of the order, will stop, waiting for other processing, this is the data processing model

Three:

Data delivery type

Of the types of data passing, there are three, which depends on the active initiator of the data passing, and the recipient of the data.

In A->b (A is active, B is taken), A->b (b is active, A is passive), a->a,a transmits the data to itself

Four:

Data incoming, outgoing, processing of data

For the data processing model we subdivide, it is divided into incoming, outgoing, data processing, three parts

4.1 Transfer of data

In the third, we see the type of data transfer, here I would like to say is the characteristics of data transmission, design to the data transfer, we need is 1. Data cannot be lost, 2 data cannot be inserted repeatedly into B,

In the above situation and requirements, we need 1: Data operation is successful, the data is marked, 2: Even after the execution, in the event of a problem, the next time, in the receiver B still have a unique field to prevent multiple insertions of data, of course, the update is generally no problem. In other words, we want to ensure that after the failure of data transmission, and so on, the data transmission of the system can still be carried out safely, ensure the data is not lost and not duplicated. That's the essence of data transmission.

V: Initiator of data transfer

There are three kinds, 1 is a remote program call, is executed, 2 is cron scheduled Tasks, 3 is the operation of the operation of customer service, regardless of what kind of operation, we have to consider the security of data, A and B are to consider the security of data, can not fully trust each other,

Remote program calls, cron scheduled tasks, usually in the form of logs, or data tables to mark each step of execution, each successful execution, will need to be marked in the database, that is, set a starting point for the next execution

, another point to note is that the data being executed, even if the next time it was executed again, or the same results as the execution, this requires the data to be executed each time the status indicator, even if the next execution, the discovery is executed, will be skipped, of course, this precautionary measure, the receiver still have to deal with, Ensure consistency of data.

For customer service operation, some operations can not be marked, after all, people can see the results, but in order to prevent misoperation, or cron, all need to ensure that the data is not lost and not be duplicated.

VI: Angle Analysis:

When it comes to angle analysis, it is the research that revolves around this thing, like Heliocentric and geocentric, from one thing to the center. Although the heliocentric represents progress, it represents only the solar system, which is not the center of the universe.

If from the data processing process, is 12345678 of the pipeline processing, but if the data from the order table of the center point of view, is that the data has been there, and then 12345678 in turn to change the data, in order

Summarize:

Involves the assembly line, General:

Total Status: Represents the result of each node processing, and the conditions under which the next node processes the data

Node state, representing a node processing data after the completion of different results, marked as different results, and then shunt, if there is only one result, then the node state is not required

Data type: This represents the source of data processed by the node, the data passed from different sources, such as the processing of the order data, the order of the consumer and the data type of the Amazon order

Operating time: This doesn't have to be told.

Operator: This is to find the corresponding person's responsibility later

The above five fields are generally necessary, each node, must have an operation time, the state of the node after operation, and the operator, each step of the node processing completed, the total state must be changed

On the development of ERP system: Data processing structure-traffic flow and data model