The bandwidth strategy is the most controversial configuration of the WiFi network's web-based system. Students feel that speed limits are too low to be used, schools feel that speed bands will be taken out by a few and IT departments are scolded at both ends.
First."Bandwidth"Think about it.
The bandwidth resources of the campus network are two-tiered: export bandwidth (total bandwidth purchased between school and operator) and inner web bandwidth (transfer capacity between campus exchange and AP). The bandwidth strategy for the campus network WiFi network, usually in charge of distribution of the export bandwidth, is not intranet bandwidth. Many people confuse these concepts, leading to a strategy that has been developed and discovered"Speed limit is not working."——It may be that there are no exits, the intranet is full or that there are no insides, limited exports but students cannot feel them.
The total amount of the export bandwidth is fixed and purchased by the school from its operator. Assuming that there is an export bandwidth for a school with 1 Gbps, 10,000 students across the school, the theoretical bandwidth for each student is 100 Kbps. But network traffic is not evenly distributed, at 8 to 11 pm it is peak time in the dormitory area, which may be three to four times the average per day, and the theoretical bandwidth available could fall to 25 to 30 Kbps——This figure is totally insufficient for the video.
So the bandwidth strategy is essentially designed to answer the question: What rules are there for distribution when export resources are not allocated? Is it first-come, no speed limit, or average (fixed speed limit per person) or priority (learning type flow over entertainment type traffic).
Three common strategy models and their application scenarios
The first is..."Hard Speed". Sets a fixed upper/down-speed ceiling for each user, such as 20 Mbps per person. The cap is the same whether it is free or crowded. This way manages simple and users experience predictable, but there is an obvious disadvantage: when they are idle (e.g. 3 am), large bandwidth is idle but each user is still confined to 20 Mbps, wasting resources.
The second is..."Dynamic Speed Limit"When network is free, users can use higher speeds; when network congestion occurs, the system automatically reduces the rate per user and ensures basic experience. This way is smarter and resource utilization is high, but complexity requires a charge system that can real-time sense export bandwidth usage and dynamic distribution strategies. Not all campus WiFi web systems support real dynamic speed limits, some are just switching between two slots and not really flexible.
The third is..."Flow Priority". Disaggregate the flow by application type——Teaching (online classroom, teaching system), academic (academic database, paper downloads), common office (mail, videoconference), entertainment (short video, games)——Different types of priority and bandwidth security are set. High-priority traffic is prioritized for congestion, with entertainment flows automatically slowing down. This approach is the most reasonable option in ideal situations, but it is highly accurate for traffic recognition, which is now increasingly difficult to identify when applied extensively to HTTPS+QUIC.
How are the parameters set: from data point of view
The bandwidth policy parameters for many schools are:"Get a head shot."This process can be significantly compressed by pre-data analysis.
Before the strategy is on, it's recommended to run for a month."Speedless."Baseline observations: Collect hourly bandwidth usage data on the export router or firewall, recording peaks, mean values and P95 bandwidth (95% bandwidth at any time). These data can show when, how much and how long the true peak is reached; what average use is for normal periods of time; whether there are unusual large traffic accounts (possibly P2P downloads or BTs).
The baseline data is available to support the bandwidth strategy parameters. For example, if peak export bandwidth accounts for more than 85 per cent of total bandwidth while online users are 2,000 during this period, the reasonable speed limit per person would be the total bandwidth.×85%÷In 2000, multiplied by a buffer factor of 0.8. This figure is then matched by mainstream video applications (about 4-8Mbps for 1080P videos), and determines whether this speed limit will seriously affect normal video play experience.
If the speed limit results are much lower than the bandwidth required for normal video play, it means that the export bandwidth itself is not enough to allow for expansion or acceptance."The peak time video experience is bad"This reality.——Strategic management can be mitigated but cannot solve the underlying problems.
The time strategy is the best way to optimize the value of sex.
In the case of limited bandwidth, time-time strategies are the most direct and effective means of optimization and cost-minimum. The basic idea is to ease restrictions during low usage (day hours, late nights) and tighten them at high usage (night accommodations).
A typical high school time strategy is probably as follows: during the day of work (8:00-22:00) each up to 5 Mbps down to 30 Mbps; at peaks (2200-24:00) every night (2 Mbps down to 10 Mbps; and at late nights (00:00-7:00) every person down to 2 Mbps to 20 Mbps.
The configuration of the time strategy is modified online, and the service cannot be restarted at any time or the ssh log server can change its configuration file.——You cannot cover the entire year with the same strategy.
Bandwidth strategy is not a one-time job.
Finally, a frequently overlooked point: bandwidth strategy is not done once and needs to be reviewed and adjusted regularly. After each term begins, the original strategic parameters may no longer apply as the number of students, type of equipment, and usages change.
It is recommended that a strategic impact assessment be conducted every semester: whether the bandwidth curves during peak times are smooth (incompromise indicates that speed does not do flat); whether there has been a decrease in the number of user complaints (also ups and suggests that experience remains poor); whether there have been significant certification failures (perhaps too low speed limits result in frequent disconnections). These indicators determine whether adjustments to the strategy parameters are needed.
A good bandwidth strategy, not finding one."Perfect numbers."The campus network is run, not done when the project is delivered.