ZDNet UK


Skip to Main Content

  1. Home
  2. News
  3. Blogs
  4. Reviews
  5. Jobs
  6. Resources
  7. Community
  8. My ZDNet

 

ZDNet UK RSS Feeds


Server platforms Toolkit

NDS tree design made easy

Ron Nutter

Published: 15 Apr 2003 13:34 BST

  • Email
  • Trackback
  • Clip Link
  • Print friendly
  • Post Comment

Coming up with a naming standard for the NDS objects
Novell provides considerable flexibility for naming NDS objects. However, if you're too detailed or descriptive in your naming, the end result could be prohibitively long. Keep in mind that the longer the names of NDS objects, the more you'll have to type when setting up logon scripts and configuring workstations to log on at the correct point within the tree, thus increasing your chances for mistakes. With all of this in mind, keep the names of NDS objects as short as possible while still being descriptive.

As you work with NetWare Administrator, you'll notice that each NDS object has its own icon that can help you identify what type of object it is. When you name objects, remember to create unique names; NetWare Administrator won't allow duplicate object names.

Even though NetWare Administrator uses a unique icon to identify an object, there are a few conventions that I try to follow when creating NDS objects to help describe the objects further:

  • PQ_ to start print queue names
  • PR_ for printer names
  • PS_ for print servers
  • FS_ for file servers

Just as there isn't one right way to design an NDS network, there's no one way to set naming conventions. Write down the reasons you've chosen to name objects the way you have. Then, if someone questions your naming conventions, you'll be able to explain your reasoning, other options you considered, and why you ruled them out.

Keeping time in sync
Just as important as properly designing your NDS is properly setting up your time synchronisation layout. Servers that are serving as Master or Read/Write replicas of a partition have to know the correct time so that when transactions arrive that affect the NDS database, such as a user logging on, the transactions are applied in the correct order -- even if they arrive out of sequence.

Time synchronisation is what distinguishes Novell's Directory Service from Microsoft's Active Directory. In Novell's Directory, the directory updates are handled in sequence by the time stamp that is a part of the record arriving at each replica server. In Active Directory, the server receiving the record assigns a serial number to it and applies it in the order in which it assigns the numbers. The time server keeps the Novell database in harmony on all of the servers in the tree.

The time server functions as an electronic clock that coordinates the time for all servers on the network. Novell uses the following hierarchy for time servers in NDS:

  1. Single reference
  2. Reference
  3. Primary
  4. Secondary

Small networks can use a single reference server on the main server in the tree as well as one or two secondary servers, which can help get the time information to the workstations and other servers on the network. The only type of server in the Novell hierarchy that I haven't actually seen in use on a NetWare network is the reference server. The reference server was initially intended to be directly attached to an atomic clock or a receiver capable of picking up WWV or WWVH, which are the two time reference servers operated on 5, 10, 15, 20, and 25 MHz by the U.S. government.

The two time server types that you'll see most often are the primary and secondary. When you have two or more primary servers, they will initiate a voting sequence between themselves to decide on the correct time and announce that time to the other servers and workstations on the network. The secondary servers will look for a primary time server to decide what the correct time should be.

The size of your tree and how many partitions you have will determine which types of time servers you'll use. One of your design criteria for the network should be to minimise the amount of traffic between sites. This criterion includes time synchronisation traffic. Therefore, I recommend at least one primary server in each partition.

Trick of the trade
Here's one trick you probably won't find in the manual or any of the TIDs on support.novell.com. When using the Configured Time Sources option in MONITOR.NLM, you need to put a :123 after the TCP/IP address or fully qualified DNS name of the time server that you're going to reference -- if it's from an external time source. If you're pointing one NetWare server to another, you don't have to use the :123; the TCP/IP address will work just fine by itself.

When you configure time servers at remote locations, make sure the server with the Read/Write replica references the primary time server at its location. By using the Configured Time Sources option in the server's MONITOR.NLM, you can point the primary time servers to each other and point the secondary time servers to the primary time servers nearest them. Setting up time synchronisation in this manner reduces the amount of service advertising that can occur when one type of time server is trying to find another type.

For a weekly round-up of the enterprise IT news, sign up for the
Enterprise newsletter.

Tell us what you think in the
Enterprise Mailroom.

Next

Previous

1 2


  • Email
  • Trackback
  • Clip Link
  • Print friendlyPrint with Konica

Did you find this article useful?
79 out of 150 people found this useful


Full Talkback thread

0 comments

Company/Topic Alerts

Create a new alert from the list below:











Microsoft Futures

Windows 7: Mixed reviews from PDC attendees

As developers received their copies of Windows 7 on Tuesday, they offered varied reactions to the Microsoft operating system update More

Microsoft floats clouds on Windows Azure

At the Professional Developers Conference, Microsoft announced the Azure Services Platform, the company's cloud-computing platform More

Ozzie: Success of Azure comes down to trust

In an interview, Ray Ozzie says businesses will be taking a risk by placing core operations in Microsoft's datacentre, but that the software giant has more to lose if things go bad More


Skip Sub Navigation Links to CNET Brand Links

Help

Become part of the ZDNet community.

Newsletters