When the university information building is up to a certain level, it encounters an inexorable topic: system integration. The residential and campus network WiFi network billing systems are independent systems that exist at both colleges, but there are natural data links between them: the boarding system knows where students live, when they leave their homes and when they enter; the billing system knows the student’s Internet account, online time, traffic consumption.
Why do you have to do this?
Let's get to the power of the docking, or we can easily make it one."It's not gonna help."The face works.
The first driver is account life cycle management. When a student is in the dormitory, the residential system creates an entry log and the corresponding network account should be activated simultaneously; when a student leaves (including graduation, transfer to a different class of accommodation, suspension etc.) the Internet account should be suspended or cancelled simultaneously. If each system is separate, there is a classic problem: a student has been out of the room for six months and his network number is running or worse.——He gave the account code to his roommate who stayed in the dorm, and the manager didn't know it was."No master account"Okay.
The second driver is network control of the accommodation dimensions. Schools may need a building that is under construction and temporarily shut down; a dormitory has been suspended for disciplinary offences; and a new-born dormitory opened two weeks before school starts, requiring a massive access to the network. These requirements are inefficient and prone to error if the information section is to operate manually.
The third driver is cost sharing and statistics. Some tertiary-level boarding networks are billed (not all of them free), with costs that may be charged to a residential bill.
Several paths to the docking scheme
There are three technical interface options, which depend on the respective openness of both systems and the IT capabilities of schools.
The first is a direct database. Each of the databases of the residential and network billing systems has a read-only account, an intermediate synchronized script, periodical rounding of the accommodation/retrieval changes to the system, and then calling the management interfaces of the network billing system for account operations. This option is simple enough that it can be done by the school IT team without two systems being in business. The disadvantage is high coupling, which upgrades could lead to script failure; real-time poor and regular rounding delayed.
The second is API docking. The residential and network billing systems expose REST APIs to data exchange via standard interfaces. This scheme is well-expanded, with both systems being upgraded independently, as long as the interface remains unchanged without affecting integration. The disadvantage is that it needs support from both manufacturers, and that there is a need for the interface document and version compatibility to be maintained.
The third is indirect docking through the Unified Identity Platform (IdP). The school ' s unified identity certification platform (e.g. OpenLDAP or Microsoft AD) manages user data for both the residential and network billing systems, both of which are in IdPira user status without directly interoperating with each other. This programme is the cleanest structure, provided that schools have established a single identity platform and both systems support matching IdP. Many schools do not have this infrastructure, so such programmes are less common in actual projects.
In practice, pits
Several problems encountered in practical projects.
The first problem is that the field is not right."Hostel ID+Bed Number"The student is marked in the web-based billing system."School number". What is the mapping relationship between them? If it was stored by a residential system, can we find out about the network billing system? If it is stored by a network billing system, can the boarding system synchronize when updating accommodation allocations? The question sounds simple, but the docking process is often stuck in the actual project because the main key design of both systems is completely different.
The second problem is the treatment of meso. Student departures are not an instant event, perhaps."I had a day off today, and I'm leaving the house."If the billing system closes the account at the first time it receives a call for departure, students will not be able to access the Internet in the last two days and there will be complaints. If they wait until they are removed, how do you know when the student actually moves?
The third problem is data consistency verification. After the two systems are connected, it cannot be assumed that the data will always be synchronized. Networks may break, scripts may go wrong, and a upgrade may have been introduced into Bug. To compare data regularly, the number of students living in the pipe system should match the number of valid accounts in the network billing system. If differences exceed 5%, the alarm must be triggered and checked manually. Without this checkup mechanism, data drift could suddenly erupt into a big problem at some point in time.
Rhythm for project advancement
The lesson of my experience is not to try to do this kind of integration in the main project. Break it down into an independent integrated module, and start the integration project after both network billing and residential systems are stable. The reason is simple: both systems will have a grinding period when they go online. If you also need to debug integration at the same time during the consolidation period, it is difficult to tell which system's pots and the processing efficiency is extremely low.
Before the integration project starts, we're gonna..."Fields and interfaces required"Make a request document that both systems' manufacturers confirm and start to develop. This document should be clear: who is the source of data, who is the consumer, how often each interface calls and overtime processing, and when error occurs. These details are not clearly written, and engineers on both sides will return to work again and again because of their lack of understanding.
And the wiring after you get online is clear: if there's a problem with the integration script, who's in charge of checking it? Is that an information division or some system manufacturer? The liability line must be written in the contract, otherwise the three parties are squeezing each other out of trouble."WiFi's broken."We're still talking about it."Whose pot is this?"。