Chapter 7

LAN System Software


Home | Chapters | Homework | Grades | Final Project | Calendar | Forums | Syllabus

Forum Policy | Assignment Policy | Games | Tests | E-Mail Mr Hull | FTP


Main | Chapter 1 | Chapter 2 | Chapter 3 | Chapter 4 | Chapter 5 | Chapter 6 | Chapter 7

Chapter 8 | Chapter 9 | Chapter 10 | Chapter 11 | Chapter 12 | Chapter 13 | Chapter 14 | Chapter 15


Summary

In this chapter we went over the different types of LAN software.  We went a little beyond the  book and talked quite a bit about single server, domain, and network directory architectures.


Networking Environments

  • Single Server: With a single server network you have to log onto the server in order to access the resources available on it.  In a small environment this works great, but once you start needing more servers it can become confusing.  Once another server is added you have to log into that server to access it's resources.  Then if you need resources on the first server you have to log in, possibly with a different password.  If the network becomes even larger you will find that the single server model won't work.

  • Domain: The next network environment is the domain.  In the domain environment you can have multiple controllers that all share the same user database.  So when you are authenticated on  the domain you have access to all resources on all servers.  (As long as your account is granted access or is a member of a group that is granted access)  The catch is only one server can write to the user database.  This server is known as the PDC (Primary Domain Controller).  All other servers in the domain have a read only copy of the database, they are called BDC (Backup Domain Controllers)  If the PDC fails you have to manually promote the BDC to a PDC.  you can only have one PDC per domain.  In a Windows NT domain structure their is a limit to the number of objects you can have in a domain.  This limit is 40,000 objects.

  • Network Directory: This third environment is designed for larger corporations.  In this environment there are multiple servers that make up a domain, but you can write to the database on each server.  This change will replicate to each other server.  In order for the change to be tracked there are two methods used.  Novell uses time stamps to determine what change is the latest.  In order for this to work all servers must have synchronized clocks.  Microsoft Windows 2000 Active Directory uses a combination of PVN's (Property Version Numbers) and USN's (Update Sequence Numbers) to determine which changes are new.  Also the KCC (Knowledge Consistency Checker) will determine the replication pattern used.

In both the Domain, and Network Directory I mentioned objects.  An object can be anything from a user to a printer.  Object are defined by their properties.  These properties can be things like a username or a printer's capabilities.  The scheme defines what properties go with the objects.  The Scheme should be expandable.  So if you wanted to add a property to a user you could.  For Example you could add a spending limit property to the user object.  Now all users will have the spending limit property.


Client Server Dialogue

We will take a look at how different client OS's know when to send an I/O request to the local computer or to a resource on the network.

  • If you are using an operating system that doesn't know about the network (i.e. DOS, Windows 3.x 9x, ME) you need to have a client for the network installed.  In this situation this client intercepts all I/O requests.  For example when you are using an application on your computer and you open a file two things might happen depending on where that file is located.  If the file is located on the local computer the client sends the I/O request to the OS.  If the file is located on a file server the client sends the I/O request to the network interface.  This client that determines where I/O requests go is called a redirector.  You can see in the image to the right how this takes place.

Operating system with Client installed

  • Alternatively you might have an OS that is designed to communicate with a network.  In this situation when you open a file the OS is the decides where to send the request.  If the file is located on the local computer the OS makes the appropriate hardware calls.   If the file is located on the network the OS send the request to the client or the redirector which in turn sends it to the network interface. 
  • Once the network interface has the request a network layer protocol data unit (PDU) is created, then it is passed to the MAC protocol layer where hardware addressing is added.  Then the request is put on the medium where the server recognizes the the request as being addressed to itself and pulls it off the medium.  On the server the request goes up the OSI layers and is answered and passed back to the client.

Operating system that recognizes the network


LAN Operating System Functions

Your LAN OS should support a few things, these included Optimized I/O, Disk Configurations and security.  Security will be talked about in another chapter.

 

Optimized I/O: LAN software should support disk caching.  This is the process of storing data in RAM so it is accessed quicker then the if it was still on the hard drive.  If data is found in memory it is called a logical read.  If it is found on the hard drive it is a physical read.
Disk Configurations: You should be able to partition the hard drive into multiple partitions. This is not required but is strongly recommended for servers.   You could have your system

Hard Drive Layout

on one partition and your data on another partition.  If your system partition becomes corrupt you will still have the data.  Also another thing you can do is keep a copy of the system in a third partition so if the main system becomes corrupt you can activate the third partition.  Partitioning gives you many options. But it is important to know what is happening on the disk.  In class I talked about a program called Partition Magic that allows you to resize and create partitions with out destroying data.  Here is a link to information on the latest version.


Fault Tolerance 

  • RAID: Redundant Array of Inexpensive Disks & Redundant Array of Independent Disks.  Even though there are many different RAID configurations this chapter only talks about RAID 0, 1 & 5. We will not talk about RAID 0 in this section because it is not true RAID.  It is not fault tolerant.  

    On a server with hard drives in a RAID configuration you will be protected in the event of a hard drive failure.  If you are using hot swappable SCSI drives you can replace the failed hard drive without shutting down the server maximizing uptime.  If you lose two drives you are out of luck.  Let's take a look at RAID levels 1 and 5.

    • RAID Level 1 is also known as mirrored disks.  There needs to be at least 2 drives in this configuration.  Identical data is written to each drive.  If one fails you have a backup.  With this configuration write times are about the same as a single drive and read times are improved.

    • RAID Level 5 is also known as striping with parity.  A minimum of three drives are needed.  The data is spread out along the drives and on each drive their is parity information written.  If one drive fails the others will be able to rebuild the information on the fly from the parity information.  In this configuration write times are about the same as one drive and read times are improved.  In the event of a drive failure performance is decreased. 

  • Duplexed Servers: In this configuration there is more then one server that are linked with a high speed connection.  The data on both servers is identical so if one dies the second one would pick up the load.

  • Redundant Power Supplies: This one is not in the book, with redundant power supplied if one fails you have a backup.  It is also good practice to have each plugged into a separate power source.

RAID Level 1

RAID Level 5


Print Spooler

Printing is one of main reasons we use networks.  In Chapter 2 we learned how to have multiple people use a printer without a network.  Now we will look at printing using a network.

When printing on a network one computer on the network would be the print server.  All other computers on the network will connect to the printer through that one computer.  The print spooler is the location were all print jobs are stored until the physical printer is ready to print.  In the diagram to the right the server would be the print server which also contains the print spooler.

Also in the diagram we see that the physical printer is attached to a print server.  Notice that the color of the text for this print server is different.  This is where things can be a little confusing.  In this case the print server provides the network interface for the physical printer.  In a network physical printer a print server is built in.

A logical printer is the printer that is installed on a server and shared out for clients to use.  A physical printer is the actually printer.  Print pooling is when you have one logical printer that is connected to 2 or more physical printers.  Print pooling is for high volume print areas.  As a side note in the Microsoft world a logical printer is called a printer and a physical printer is called a print device.


More Information

  • This site breaks down all the RAID levels along with diagrams and list's the advantages and disadvantages of each level.

  • Microsoft's Active Directory page.

  • Novell's NDS page.

Click Here to download the slides for this chapter

(NOTE: You must have PowerPoint or PowerPoint Viewer if you don't have  either Click Here to download PowerPoint Viewer.)


Home | Chapters | Homework | Grades | Final Project | Calendar | Forums | Syllabus

Forum Policy | Assignment Policy | Games | Tests | E-Mail Mr Hull | FTP