The system usually goes well on the day it gets online, and the company’s technical support is still there, and its colleagues are watching. But the real test is three to six months after it has been online, and the vendor enters the normal mode, and the student use starts increasing, so that problems that have not occurred in the testing environment continue to emerge.
A classic problem of authentication that is successful but not connected to the network
Student response"The sweep has passed, but the net still doesn't work.", the largest number of complaints during the period. The root cause of these problems is not usually the certification system itself but that the authentication status has not been correctly sent to the access device (AC or AP). When there is an interactive timeout between the authentication server and the access device RADIUS or Portal, the server records this side of the document."Authenticated", but the access device is not yet a strategy for open users. Students can usually be connected again or disconnected, but the user senses it is."There's a system problem.". The problem is more likely to occur in a high-situation scenario.——At peaks, such as the start of school and before vacation, emphasis is placed on the response time under the access strategy rather than just the logbooks of the authentication server.
Billing data are not consistent with students' actual perception
Another type of HF complaint during the period of operation is"It's not working, but the traffic is out.". The following are the main directions: first, multiple equipment-sets, student cell phones with hot spots and roommate traffic on the same account; second, silent flow from backstage applications, especially from Android, which is more than expected in some settings; third, system measurement accuracy, where there are mass delays in measuring flows, resulting in discrepancies between the figures and real-time residuals shown in bills and causing students to suffer from a significantly higher consumption of traffic in certain settings."Just a moment of light.". The check-up is done separately: the logs are kept up and down, raw data from traffic measurement services, processing times for bill generation services, and the logs of the three links are looked after, not just bill interfaces.
Data quality problems imported by batch accounts
The primary task of the information centre is to start school every year, import new accounts from the teaching system. If data imports are not properly formatted, there will be a problem with account states: some study numbers contain special character import failures, some name fields contain spaces that cause authentication failure, and others have cell phone numbers which are not tied. Each of these problems is small, but if there are 500 in 10,000 data imported, the information centre will manually check 500 sheets. This is done by adding a layer of data pre-check before importing processes: running the source data through the format check scripts, listing problematic entries for manual processing instead of checking them after importing, etc.
Performance degradation during seasonal peak periods
Many schools have a well-established Internet billing system with daily loads, but this seasonal peak has triggered slow response, Portal page Carton, and partial overtime requests for certification. The reason is that the system does not usually reserve extensions: databases are full, billing lines are inadequate, or bandwidth of the Portal server is squeezed. The solution is either designed at maximum capacity during deployment or a flexible extension mechanism (which is less easily achieved in the high school self-built housing environment). The most realistic approach is to pre-screen the resource utilization of system services one week before each peak period, and to extend or adjust parameters for components that are close to the ceiling.
The personnel handover problem due to the missing transport document
The mobility of technical staff in information centres is a common problem at universities, and many of the system's dimensions may exist only after the departure or rotation of engineers responsible for Internet billing systems"The last one knew."state: where the password is written, on which server the configuration file is placed and why an unusual parameter is configured this way……This knowledge is not captured into a file, and each time the person turns up is re-explored and often becomes aware of the document missing after problems have arisen. It is proposed that in the first year of operation of the system, one to two weeks be devoted to writing the manual: server lists, key configuration instructions, frequently asked questions, and how the manufacturer will contact it.
Vendor response quality decay
This is a problem that many schools have a sense of well-being but rarely pay attention to in evaluating bids: the vendor ' s technical support response was timely during the project ' s online phase; the time frame and quality of response will decline significantly after entering warranty or after-sale service."Submission of worksheets"The work order came in for days, and then they said,"Need next update to solve"This is common in projects where there are no explicit SLA terms in the contract. It is proposed that post-sale response periods be included during the negotiation phase of the contract: 48 hours for general questions, 4 hours for issues affecting normal use, 1 hour for core functions (unable to authenticate, billing stops) and a default clause attached.
Reconciliation of the operator ' s settlement cycle with bills
Some of the school-to-work system's flow costs involve settlements between schools and operators, such as a part of the traffic packages that students buy is an export flow from the operator. In this scenario, schools are required to check their bills regularly with the operator to confirm consistency between the student purchase volume and the operator measure. If the calculator on both sides (e.g. schools measured at TCP level and operators measured at IP level) there are differences in billing, which may involve significant amounts after accumulation over time. This issue does not necessarily occur in every school, depending on the costing model, but often has been backloged for several months at times, and reconciliation is cumbersome.
The problems of the operational period are not largely technically unsettled, but rather are less efficient than pre-empting them at the selection and uplink stages, when they arise. Taking these types of common problems into account in the system building phase would be much less effective than reactive responses after going online.