Printed Study Guide from http://www.Cramsession.com

IntranetWare 4.11 Design and Implementation

Project Approach Phase

1. Assemble the project team
2. Gather information about the company and their network

Gather company organization charts, WAN and LAN details charts, and information about all network resources
Gather workflow information - Where the company is managed and how various tasks are performed

3. Train the project team
4. Define the project scope

Which team members will be needed to complete the project?
Will time synchronization planning be needed?
Will you need to use partitions and replicas?
Will you need additional servers and workstations?

5. Create the project design schedule

Project Team

IS Manager	Coordinates with NDS expert Oversees design Coordinates implementation Manages cost and time estimates
NDS Expert	Project lead for design process Creates directory tree design Plans NDS security, partitions and replicas, and time synchronization Coordinates login scripts with team Ensures design is documented
Server Specialist	Plans pilot installation Maintains performance levels Implements server upgrades and migrations Creates protocol standards Monitors migration Plans disaster recovery procedures
Workstation Specialist	Upgrades client workstations Determines workstation hardware requirements and compatibilities Monitors client performance Designs login scripts; coordinates with team
Application Specialist	Maintains, upgrades and migrates server, client and network applications Ensures stability of applications
Printer Specialist	Installs and configuring printers Upgrading client workstation's drivers Ensuring users can access printers Handles parts of migration that involve printers
Connectivity Specialist	Handles physical aspects of network Manages protocols, and maintains LAN and WAN efficiency
Testing Lab Coordinator (Used in large scale design)	Evaluates and tests clients, servers, and software prior to implementation Runs diagnostics on equipment and gathers information needed for test
Education and Training Coordinator	Trains team members and/or users in unfamiliar subjects

Designing the Directory Tree Phase

1. Creating standards - Includes naming standards container and leaf objects, and property lists for each object.

2. Organizing upper layers of the tree

Organize by location - Used to divide organizations separated by WANs. Reflects WAN topology and reduces WAN traffic.
Organize by function - Used to divide organizations by departments and divisions. Makes partitioning difficult. Used mainly when organization is in one location or if WAN links are extremely fast and reliable.

3. Organizing lower layers of the tree - Can be divided any way that provides a benefit without causing network problems. Can be strictly functional. To determine the structure of lower levels, consider which users will need access to the same resources.

4. Modifying Your Design

Administration

Centralized - Uses a single administrator/department to control all objects in the directory tree.
Server container - Option used with centralized administration; places all servers in a single container object to simplify administration and enhance security.
Distributed - Uses separate administrators for portions of the directory tree.

Partitioning and replicas - Objects connected via WAN links should not be in the same partition, which increases traffic and errors, and slows synchronization.

Bindery services - NetWare 4 allows up to 16 bindery services on a single server. If over there are over 16 contexts, consider combining some of the containers.

Login scripts - Combine departments into a single OU that have similar login script requirements. Consider subdividing an OU that requires two or more login scripts.

Determining Partition and Replication Strategies

Name Resolution - Process of locating an object on the network using distinguished names. First looks at the server you are attached to, then follows subordinate references to find the object. Subordinate references across a WAN link slow the process. It is best to place a replica on a local server for each container that will be accessed frequently by that server.

Partitioning considerations

Default scheme is sufficient if:
      1) Number of servers used for replicas is 15 or less
      2) There are no WAN links to other locations
      3) there are a total of 5,000 or less objects

Custom scheme:
      1) WAN links: Keep each location in a separate partition
      2) Location: Group replica servers that are near each other in a single partition
      3) Number of Replicas: When more than 15 servers have a replica of a partition, split the partition or eliminate the number of replicas
      4) Number of Partitions: Carefully consider the number of partitions to use, as too few or too many can cause problems.

Rules for partitioning

Partitions can not overlap. An object can only be in one partition.
Partitions affect NDS data only. Does not affect file system.
Partition can only be stored 4.x servers, and not on 3.x or earlier servers.
Empty container can not be used as a basis of a partition. Must contain leaf or container.
Partitions are named after the topmost container object.
Directory tree should be divided into only a few partitions at the top level and more partitions at the lower level if needed.

Replica placement guidelines

Create at least 3 replicas for each partition
Be certain you have replicated the [Root] partition
Create replicas close to the users who use objects in that partition
Bindery services require a read/write replica of the partition containing the bindery context.
Do not rely on subordinate references for fault tolerance.
Keep one replica at a separate location for fault tolerance.

Replica Types

Master

The main copy of a partition. Only one master replica is allowed per partition. Receives update requests from Read/Write partitions, and pushes synchronization updates to all other partition types.

Read/Write	Copy of a master partition. Can be read from and written to. Sends all change requests to the master replica of a partition. Receives partition synchronization updates from the master replica.
Read	Copy of a master partition. Can only be read from. Are used for searching and viewing objects only. Receive synchronization changes from master and read/write replicas.
Subordinate Reference	Special replica pointer. Used to redirect NDS queries to the proper replica. These are created automatically on a server when it contains a parent replica, but no replicas of the partitions of that parent's children. These cannot support object or partition changes.