Businesses usually watch product introductions, listen to suppliers and compare price of parameters when choosing wireless certification systems, but one thing many people don't do.——POC test.
PoC is Proof of Concept, conceptual verification, and you find a supplier to run around the real environment to see how the system works. And here's what POC should focus on at this stage, which pits are not visible by parameters.
It's a long, long time ago.
It's in the product manual."Support 5000-ming."And that's the theory peak. The real scene is very different from the lab data.
Measurement: Search for a real network environment, simulation of peak scenarios such as 9 to 10 a.m. when everyone is online, or before and after the full meeting. Filling staff up, depending on how long the certification system responds under this pressure, will be an increase in the failure rate of accreditation, slow response, and offline.
I saw a supplier's product that ran well in the testing environment, and when it reached the client site to connect to the real network, 200 cards were sent and started. Then they found out that their test environment was using gigabytes of inner grids, with the client site network architecture gathering 100 trillions of dollars, and the bottlenecks were not on the certification system itself, but rather that the bottom design was irrational.
Are we connected to the existing network?
This is the most problematic part. It says in the product manual:"Support H3C.", however, the model and solid version of each manufacturer varied widely and the results of the docking varied considerably.
(c) Measurement: send specific models and solid version of existing network equipment to the supplier to confirm that it is combing."I think so."It is important to be on the spot and not to believe in oral responses.
One customer has stepped on the pit: they use a Chinese AC, and the supplier says it supports it. It's not right to go online, but it needs to upgrade the AC solid. But upgrading the AC solid involves business interruption, scheduling the window period, and it takes a month to get it done.
Will the network break when it's broken?
Some enterprises have a network-based certification system, which is not accessible to all once the certification system has been shut down. This situation needs to be clear at the selection stage and use sidewalking mode when deployed.
Measurement: The testing phase stops the authentication system process to see if the network is broken. If the network breaks, it indicates a serial deployment with a single point failure risk. If the network runs as usual, it means that the system does not affect network availability.
This is particularly important for the production environment.
Can the account be synchronized?
If an enterprise has a AD domain or HR system, the account synchronisation function is to be measured once and not only by a demonstration.
Measurement: Simulates the entry of employees in a test environment, creating accounts on the AD side to see if and how long it takes to automatically synchronize with the authentication side. Simulation deactivates the AD side account number to see whether the side of the certification system is synchronized or not, and if an online terminal will be kicked offline.
This function usually performs well in the presentation environment because of the small amount of data available and the smooth flow. In a real environment, however, if AD is more complex in structure or there is a data mapping logic between HR systems and authentication systems, the probability of stepping on a pit is not low.
Daily transport is not going well.
The POC phase is best to involve those who are really responsible for the carrier in the testing, not as a pre-sale consultant demonstration but as an operational user.
The measurement: a few specific transport scenarios are proposed, depending on whether they can be quickly completed in the backstage. For example: batch import of 50 new employee accounts; process a password-forgotten self-help recovery process; check an employee's Internet log for this month; and set a visitor account for 72 hours.
These operations appear to be simple in the presentation environment, but when account numbers are large or involve multiple-level access management, backstage operational response speeds vary significantly.
How's the mobile end?
Mobile phones, tablets, and so on account of the mobile terminal. Portal certification pages show off on a cell phone, clean up authentication processes are not smooth, mobile accounts are tied to steady and unstable levels, all of which are measured realistically.
Measurement: Run the full authentication process with its own mobile phone and several types of machines common to companies. These include: connecting WiFi to pop-up Portal pages, entering account codes, submitting authentications, validating successful jumps, each step being taken to see if Caden is there or any unusual display.
I've dealt with a case where, after some Andre phone connected to WiFi, Portal's page came out, but the keyboard that entered the box on the page blocked the confirmation button and the key was filled in."Login"But it's blocked by a keyboard, which is difficult to spot. It's not easy to find in the demonstration environment, but it needs to be measured using real machines.
We'll do more POC tests in the selection phase and less pedals after we get online.
Some companies feel that the opt-in phase is wasting time with POC and want to sign contracts quickly. But, from actual project experience, those that have done a full POC are much more satisfied with being online than they have not been.
The problem was identified during the selection phase, where costs were borne by suppliers; it was discovered only after being online that costs were self-inflicted.