Many units are on the Portal certification system, and the first reaction is to select a product that can pop the log pages, and then it turns out that the page is just the outermost level, with two things left open, and the next day there are holes. One is a way of authentication, using account codes, text messages, micro-letters or decertification; one is an account number system, where people come from, who has access, how long their life cycle is. Without being able to sort them out, the more difficult they use the system, the less it can be said, the safer it can't be found, even if it can't explain who they are online.
The way you authenticate it is not to be a one-size-fits-all.
The way the field uses a code for accounts plus a domain account is the most secure, public area has been either text or micromail sweeps, and partners are the least careful with temporary visitors. One way to get out of the world is either bad or safe. The choice also needs to be end: mobile phone users can clean up their hands, old equipment may only support an account password, leaving room for one. The most direct price of wrong selection is complaints and bypasses, and some people borrow another account number to use the Internet, which causes them to mess up and break security from the bottom.
The account system first has to figure out the source.
Where accounts come from, they are built manually, synchronized from membership systems or match third-party sources, determining whether the cost of back-carriage is high. Manually constructed for small sites, thousands of units are built manually, and there are ready-to-use systems; but the matching source is reliable. If the source is not clear, the account number becomes unsourced, missing, unrequited.
Life cycle is more important than building a account.
The accounts are not built and the new ones are effective, the checkbacks, the freezes, the alarms are out. Many people just look at how to open them, forget how to close them. If a non-recoverable account is a security risk, the separating person can still access the Internet.
Permissions and roles are designed separately.
The difference between the two, which is what the role can do and where it can be done. A mix of the entire configuration is affected. After separation, changes in roles are made to the authority, without moving the power structure, so that organizational changes do not re-create the system. Permissions also have to drill into regions and time: a number only in the office area, only work is effective, rules are detailed, security and experience are balanced, and it facilitates an ex post facto review of the basis for authorization of a given access.
Identification and de-certification areas to be cleared
Not all places are certified. The white list may be exempted from the airscreening of the conference rooms, device debugging, and the physical network terminals. Which areas are certified, which are cleared, and which are clearly drawn.
Account name has to have rules that can be retrieved
The name is not to be named with a bunch of meaningless numbers, but it can be attributed to: department, type, validity. The naming code allows you to identify who at first sight. The name is confused and the draining takes half a day to find someone in the account, which is delayed by response time. It also avoids conflict and speculation: do not use simple rules to allow outsiders to export another account, given that it requires both information volume and travel-proof balance between security and availability.
Leave a list of acceptables before you get on the line.
After both things were fixed, a list of the way in which the account was authenticated, the source and synchronised, the life cycle rules, the permission role matrix, and the exemption white list were entered into. One article passed, and none of them was matched back to back. The list was also a prototype for the handover document, with the change of owner being matched. It was not written out of lock drawers, but it was updated with each rule that made it viable; otherwise no one would have said why it was so good six months later.
Do a greyscale check before getting online
Two things are set and then cut in full, so take a small group of real users for one week to see if the authentication is not working well, the account is synchronized or not. The grey phase is the least costly and the whole amount is online before it affects everyone. The gray scale focuses on three categories: whether the new account is alive, whether the check-out account is frozen in time, and whether the certified white list is off-limits. All three groups have steady re-introduction, so they have a bottom, and they come online without saving fire.
The executive team needs to eat the scene before it does.
The team will have to leave a record of the decision: why this authentication method is selected, why the account source is so connected, and why it is followed by the project, and when there are other people or disks behind, it is not possible for anyone to say what they thought after six months.
The most time-consuming thing for Portal certification is not to do an interface, but to align the authentication method with the account number system. It's a choice, source clear, life cycle closed and permission role separate, and the page is a good one; these two pieces are confused and a nice logbook will not save a single puzzle.