Discussion Locked

NDS Design Opinion

0 helpful
9 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

I am involved in a debate on directory design.

Which is better for creation of a tree for a small company with multiple WAN linked offices, some limited job role differentiation and little need for security other than privacy for home directories and personnel records?

1. Creation of multiple trees consisting of a single container holding all objects, along with a single map to root of the volume and use of groups and inherited rights filters to restrict/allow access to folders in volumeincluding home directories and sensitive personnel information (personnel records used by an accounting/personel office in one office).

or

2. Eventual creation of a single tree with duplicate configurations for each office in which multiple continers relating to various job junctions are created below the container holding server, printer, volume email and other "network" objects, creation of a second volume on the server containing personnel information and the use of container rights and login scripts to provide access to various volume and/or folder configurations.

Which in you opinion is the best most effecient strategy?

9 replies

by kevin anderson

Both.

#1 Will work best if you can have administration at each of the WAN locations, and a very small Corporate level admin overseeing everything.

#2 will work best if you want to control everything centrally, rather than having each group install their own patches, etc.

From a user perspective, they will probably prefer #1, since their requests (please add a new userID for the the new person who arrived 10 minuites ago and is waiting to log in right here...) will be responded to much faster. Additionally, if there is ever a serious NDS problem, it will probably only affect one WAN link, rather than the whole group.

Corporate Admins though, will prefer option #2, since it forces a more team oriented approach to administration. Rogue admins would be unlikely (or unable) to patch a server without everyone being made aware in advance... And company-wide changes would be much easier to enact.

If you have central management, go with #1.
If you have distributed management, go with #2.

Kev.

by acurrin

If there is any need for access to files across the WAN between offices, it will be far better from and Admin and user perspective to go with one tree. You would want a partition of the tree at each site to reduce replication traffic and replicas oneach server to service authentication if the WAN link is down.

The single identity of each individual on the network will make managing access to remote resources far easier, as well as the movement of users between offices.

by bdgray

Single tree with branches(OUs) for each remote site. You do not want NDS to have to tree walk accross a WAN link so have a primary server for each locationv and partition NDS around those locations.Place R/W replicas of the other NDS partitions on each server in each location.

by estebandelatorre

Follow this "Practical rules"
1.- Single Tree -> Is a Single company!
2.- One container per Site? Only if there will be a server (e.g. for home directories)in each site and at least 10 users.
3.- Single container if the ?n? servers are
ALL in the same place (up to 300 objects ok)
4.- Separate container for the roles (enhaced admin rights)

good luck!

edt@bancobgn.com

This discussion is closed to new replies.