Especially in the case of wireless accommodation, high-sum certification, multiple exports and multi-operator co-existence, the design level of a billing engine directly determines that the system is“Steady.”Or...“Avalanche”。
The most recent project in the university is that it has been implemented for more than three years.
The centralized network break-up, 99 percent of which occurred in the billing system, not in the export chain itself.
The cloud-based campus network Authentication & Billing system, which is at the cost engine level, has eliminated this risk early.
One, a centralized network is essentially a costing engine.“Misjudged state.”
In the NATSHEL_AUTH_BILLING system, centralized network breakouts typically result from a number of costing logic problems in local deployments or early design:
-
To judge a short network shaking as a true decoupling of the user
-
The export is an anomaly that equates to the user's downline.
-
Billing sessions and access sessions are tied to the
-
Costing status stored in a single node
Once any of these items are triggered at the peak of the dormitories, they will appear:
-
A lot of users are being used simultaneously.“Clear Underline”
-
Student recertification failed
-
Present.“The entire floor is off the grid.”Phenomenal
These are not problems that are not functionally inadequate, but rather...Structural problems at the design level of fee-for-engine。
II. Cloud deployment first changes“Position of the billing engine”
In the cloud-based campus network Authentication & Billing, the billing engine is no longer dependent on a local device but has the following characteristics:
-
Run independently of export equipment
-
Independent of access control logic
-
Maintain user billing status independently
The point of this step is:
Even if there were a vibration in the equipment exported, it would not directly affect the judgement of the billing engine.
This is the first line of defense to avoid a centralized network breakout.
III. How to distinguish cloud-based engines“Unusual break”and“Real offline.”
In the real tertiary environment, the following are occurring every day:
-
Wireless AP reselect
-
Terminal toggle frequency
-
NAT session is briefly interrupted
-
The exit link is shaking.
If the billing engine is in these circumstances,“Clear it now.”In the end, a centralized network is almost inevitable.
A mature campus network Authentication & Billing system, which is usually introduced in cloud-end billing engines:
-
Multi-state judgement mechanism
User status is not simple“Online/ Offline”But:-
Online
-
Suspicious break.
-
Waiting for confirmation
-
Real offline.
-
-
Time buffer window
The billing engine does not immediately end the session because of a link anomaly, but enters the observation window. -
Multiple-sign cross-checking
Combining:-
Heart rate
-
Changes in flows
-
Access Side Status
-
Historical behavioural models
-
Only when multiple conditions are met will users be judged to have actually left the line.
This logic is almost done only on cloudy ends.
IV. The billing and access sessions are thoroughly uncoupled, key to avoiding avalanches
In the web of easily problematic campuses, Authentication & Billing systems, common designs are:
Access cut out = billing over
The core design of the cloud-based campus network Authentication & Billing is:
Access Session Change ≠ End of the billing session
This is reflected in:
-
Users briefly offline, billing sessions remain
-
Export switch, no triggering fee settlement
-
AP roaming without affecting billing
The direct result of this is:
-
It won't be because“Local problems”Activate“System-level offline”
-
The billing engine does not have a chain reaction during peak periods.
V. How do cloud-based engines respond?“Focus on the tide.”
The concentration of the net is often accompanied by a second risk:
Focus reconnect.
In local billing systems, the following are common:
-
A mistake to get off the line.
-
Thousands of users re-certification at the same time
-
Authentication queue stack
-
The system is out of control.
The cloud-based campus network Authentication & Billing system will do two things ahead of schedule at the billing engine level:
-
Limit settlement trigger frequency
Prevent large numbers of users from being cleared simultaneously in a short time. -
Authentication and billing error
Even if a reconnection occurs, it will not trigger the reconstruction at the same time.
This keeps the system in an unusual situation.“Controllable degradation”Not“It just collapsed.”。
VI. FREEDOM OF INTELLIGENCE: HOUSE LOCKING“Networking”Key changes
Some undergraduate institution (faceless):
-
Number of students 20,000+
-
The dorms are all wireless.
-
Export by three operators
Early system issues:
-
Late, 8 o'clock.
-
Students complain frequently“I got kicked right on the line.”
-
The transports cannot recover the problem.
Switch to the cloud-based campus network Authentication & Billing system:
-
Billing engine running independently
-
Session status cloud maintenance
-
Export abnormally untriggered settlement
The results were:
-
Hostel peaks are running steadily
-
No more centralized silos.
-
There has been a marked decrease in the number of complaints
It's not a network.“It's better.”But...The billing engine is not to hurt the user.。
VII. NatShell Core trade-off on the design of a billing engine
NATSHEL_BRAND has been following several principles in the design of a billing engine for the Authentication & Billing system on campus:
-
The billing state must be cloudy.
-
Do not treat any single device as“Final decision”
-
Not for short-term abnormal settlement of users
-
Priority is given to ensuring overall stability of the system
This is why the system can be stable in a long-cycle, high-separation accommodation environment rather than being driven by manual effort.
VIII. Just say one word from a product perspective
In the university environment:
The centralization of the network has never been an accident.
It is the logical consequence of the maturity of the design of the billing engine.
The cloud-based campus network Authentication & Billing was designed to absorb this risk early through the management of billing engine status, session decoupling and abnormal absorption.