CHAPTER 5

ADVANCED FILE SYSTEMS

This chapter introduces the distributed file system (DFs) and the File Replication Service (FRS).

Dfs allows system administrators to make it easier for users to access and manage files that are

physically distributed across a network. Users no longer need to know and specify the actual

physical location of files in order to access them.

Lesson 1: Distributed File System

Dfs for Windows 2000 Server provides users with convenient access to shared folders that are

distributed throughout a network. A single Dfs shared folder serves as an access print to other

shared folders in the network.

Dfs Overview

Dfs is a single, logical, hierarchical file system. It organizes shared folders on different computers

in a network to provide a logical tree structure for file system resources. Dfs can organize

resources that reside on different components of a network.

Because the Dfs tree is a single point of reference, regardless of the actual location of the

underlying resources, users can easily gain access to network resources.

A user who navigates a Dfs-managed shared folder does not need to know the name of the

server on which the folder is shared. After connecting to Dfs root, users can browse and gain

access to all resources below the root, regardless of the location of the server on which the

resource resides.

A Dfs share uses a tree structure containing a root and Dfs links.

*** See the chart on page 215-216 ****

Types of Dfs Roots

The Dfs Service is auto-installed with the installation of Windows 2000 Server. The service

can be paused, stopped, and started but not removed from the operating system.

Two types of Dfs roots can be configured on Windows 2000 Servers: Stand-alone Dfs roots

and domain Dfs roots (sometimes called fault tolerant Dfs roots).

======================================================================

winser5.html PAGE 2 2002/01/05

Stand-Alone Dfs Roots

Below is a list of common characteristics for Stand-Alone Dfs:

• Stand-alone Dfs information is stored in the local registry.
• A stand-alone Dfs root permits a single level of Dfs links.
• When using the Distributed File System snap-in to connect to existing stand-alone Dfs roots, all

servers known to the browse list are retrieved since there is no unique NetBIOS name registered

by Dfs-enabled servers.

• Stand-alone Dfs roots can be located on all supported file systems, although locating resources

on NTFS formatted partitions is recommended.

• Stand-alone Dfs roots offer no replication or backup; consequently, the Dfs root represents a

single point of failure. You can create a replica from a stand-alone Dfs link; however, file

replication services are not available.

Domain Dfs Roots

Below is a list of the common characteristics to Dfs fault-tolerant roots:

• In a domain Dfs root, multiple servers hand out referrals for the Dfs namespace. Fault tolerant Dfs

roots use Active Directory services to store Dfs tree topology and remove the root as a single point

of failure.

• A fault tolerant Dfs root is stored in Active Directory services and replicated to every participating
• Dfs root server. Changes to a Dfs tree are automatically synchronized with Active Directory

services.

• You can implement fault tolerance at the file and content level by assigning alternate resources to

a Dfs volume.

• Fault-tolerant roots must be located on NTFS version 5.0 formatted partitions.
• The list of domains and servers is populated by querying the global catalog for all fault-tolerant

Dfs roots

• (ObjectClass = ftDfs)
• Dfs replication topology uses the existing Active Directory replication topology.

Configuring Dfs

Windows 2000 allows you to configure stand-alone Dfs roots, Dfs links and domain Dfs roots.

=====================================================================

winser5.html PAGE 3 2002/01/05

Configuring a Stand-Alone DFs Root

Stand-alone Dfs stores the Dfs topology on a single computer. This type of Dfs provides no fault

tolerance if the computer that stores the Dfs topology or any of the shared folders that Dfs uses fails.

A stand-alone Dfs root is physically located on the server users initially connect to. The first step in

setting up stand-alone Dfs is to create the Dfs root.

To create a stand-alone Dfs root, use the Distributed File System snap-in to start the New Dfs Root

wizard.

The following table describes the screens within the wizard and what actions you can take to configure

the new Dfs root:

======================================================================

Screen Actions

======================================================================

Select The Dfs Root Type Select the Create A Stand-Alone Dfs Root

Option

Specify The Host Server Enter the initial connection point for all

For the Dfs Root resources in the Dfs tree. You can create a

Dfs root on any computer running Windows

2000 Server.

Specify the DFS Share Enter a shared folder to host the Dfs root.

You can choose an existing shared folder or

Create a new share.

Name the Dfs Root Enter a descriptive name in the Comment text

Box for the Dfs root.

Completing the New Dfs Root Review the settings for the Host Server, Root

Wizard Share, and Root Name text boxes.

=====================================================================

Configuring a Domain Dfs Root

Domain Dfs writes the Dfs topology to the Active Directory store. This type of Dfs allows Dfs links

to point to multiple identical shared folders (also called replicas) for fault tolerance.

To create a fault-tolerant Dfs root, use the Distributed File System tool to start the new Dfs Root

Wizard.

=====================================================================

winser5.html PAGE 4 2002/01/05

Configuring New Dfs Links

Users can browse folders under a Dfs root without knowing where the referenced resources are

physically located. After you create a Dfs root, you can create Dfs links (also known as child nodes).

To create a Dfs link, open the Distributed File System snap-in, and click the Dfs root to which you will

attach a Dfs link.

The following table describes the option in the dialog box:

======================================================================

Option Description

======================================================================

Link Name The name below the Dfs root that users will see when they

connect to Dfs.

Send the user The UNC name for the actual location of the shared folder

to this shared the Dfs link refers to.

folder

Comment Additional information (optional)

Clients cache Length of time for which clients cache a referral to a Dfs link.

This Dfs After the referral time expires, a client queries the Dfs server

Referral for about the location of the dfs link.

X seconds

======================================================================

Dfs link will appear below the Dfs root volume in the Distributed File System tool and will appear

to a Dfs enabled client as a folder below the Dfs root.

Lesson Summary:

• Dfs provides a convenient way for users to access shared folders that are distributed throughout a

network. A single Dfs shared folder, called a Dfs root, serves as an access point to other shared

folders in the network, called Dfs links.

=====================================================================

winser5.html PAGE 5 2002/01/05

• Dfs organized shared folders on different computers in a network into a single, logical,

hierarchical file system.

• Two types of Dfs roots can be configured on Windows 2000 Servers: stand-alone Dfs roots

and domain Drs roots.

• Stand-alone Dfs stores the Dfs topology on a single computer. This type of Dfs provides

no fault tolerance if the computer that stores the Dfs topology or any of the shared folders

that Dfs uses fail.

• Domain Dfs writes the Dfs topology to the Active Directory store. This type of Dfs allows
• Dfs links to point to multiple identical shared folders and supports file replication for fault

tolerance.

• When changes are made to a Dfs link that is part of a domain Dfs root, the changes are

automatically replicated to other replica members.

Lesson 2: File Replication Service

FRS is the file replication service in Windows 2000 Server. It is used copy and maintain files on

multiple servers simultaneously and to replicate the Windows 2000 system volume (SYSVOL)

on all domain controllers.

FRS Replication ***On test ***

FRS is installed automatically on all Windows 2000 Servers. It is configured to start automatically

on all domain controllers and manually on all standalone and member servers. Although Active

Directory replication and the FRS are independent of each other, they share a common replication

topology, terminology, and methodology. Active Directory store uses FRS to synchronize the

directory among all domain controllers.

Each Windows 2000 domain has one or more servers that serve as domain controllers. Each

domain controller stores a complete copy of Active Directory store for its domain and is involved

in managing changes and updates to the directory.

Within a site, Active Directory services automatically generates a ring topology for replication

among domain controllers in the same domain.

The ring structure ensures that there are at least two replication paths from one domain controller

to another; if one domain controller is down temporarily, replication still continues to all other

domain controllers.

Active Directory services uses multimaster replication, in which no one domain controller is the

master; instead, all domain controllers within a domain are equivalent.

=====================================================================

winser5.html PAGE 6 2002/01/05

If you add or remove a domain controller from the network or a site, Active Directory Services

reconfigures the topology to reflect the change.

Sites and Replication

A site is made up of one or more IP subnets that identify a group of well-connected computers.

Only those subnets that share fast and reliable network connections of least 512 kilobits per

second (Kbps) should be combined.

Domain structure and site structure are maintained separately in Active Directory services. A

single domain can include multiple sites, and a single site can include multiple domains or parts

of multiple domains. See page 230.

There are two types of replication: intra-site replication and inter-site replication:

Intra-Site Replication *** ON TEST ***

Intra-site replication has the following characteristics:

• Intra-site replication occurs between domain controllers within a site.
• Replicated data is not compressed.
• The default replication interval is 5 minutes.
• Replication is trigger-based, that is, notification and pull.

Inter-Site Replication

Inter-site replication has the following characteristics:

• Inter-site replication occurs between domain controllers in different sites.
• You can specify the time when inter-site replication should occur. The

default replication interval is 3 hours.

• You can specify the network transport for use in inter-site replication.
• Inter-site replication is compressed, regardless of the transport used.
• Inter-site replication compression reduces the data on the network by 88 – 90%.

Knowledge Consistency Checker ** ON TEST **

Within a site, a process called the Knowledge Consistency Checker (KCC) generates a ring

topology for replication among domain controllers in the same domain.

=====================================================================

winser5.html PAGE 7 2002/01/05

This ring structure guarantees that there are at least two replication paths from on domain

controller to another, ensuring that if one domain controller is temporarily down, replication

will continue to all other domain controllers. In addition, the ring structure is created such that

an update takes, at most, three “hops” from the domain controller where it originates to any

other domain controller in the site.

The KCC periodically analyzes the replication topology within a site to ensure that the replication

topology is efficient.

NOTE: Administrators can make modifications to the replication topology, including changing the

schedule for inter-site replication, to meet the requirements of an organization.

Unique Sequence Numbers

When a directory object is updated at a domain controller, either through a change that a user

or administrator makes or by replication from another domain controller, the domain controller

assigns the change a Unique Sequence Number (USN).

Each domain controller maintains its own USNs and applies USNs incrementally to e

each directory change made at the domain controller.

When the domain controller writes the change into the directory, it also writes the USN of the

change with the property.

Each domain controller maintains a table of the USNs that it receives form every other domain

controller in the domain, and the table lists the highest USN that is received from each domain

controller. Each domain controller then periodically notifies the other domain controllers in the

domain that it has received changes and sends its current USN. If there are changes and the

domain controller has not received them, it requests that only the changes be sent.

Using USNs eliminates the need for precise timestamps for changes and for time to be synchronized

precisely among domain controllers within a domain.

The use of USNs also simplifies recovery after a failure. When a domain controller is running

again after a failure, it restarts replication by asking each of the other domain controllers for

changed USNs greater than the last USN in the tale for the domain controller. Because the

table is updated automatically as the change is applied, interrupted replication cycles pick up

exactly where they left off, with no loss or duplication of updates.

=====================================================================

winser5.html PAGE 8 2002/01/05

Implementing FRS

Implementation of FRS consists of several phases: replicating SYSVOL, replicating domain

Dfs roots, and configuring FRS for inter-site replication.

Replicating SYSVOL

Changes to the %systemroot%\SYSVOL directory on any domain controller are automatically

replicated to other domain controllers within the site. The replication topology and process is

separate but identical to Active Directory replication. There is automatic replication to the

%systemroot%\SYSVOL folder, and all domain controllers are updated.

The default folder structure is as follows:

• %systemroot%\SYSVOL\Sysvol\domain_name\Policies
• %systemroot%\SYSVOL\Sysvol\domain_name\Scripts

Any files and folders added to %systemroot%\SYSVOL\Sysvol\domain_name are automatically

replicated.

Replicating Dfs Fualt Tolerant Roots

Dfs uses FRS to replicate data in domain Dfs links. When changes are made to a domain

Dfs link that is part of a domain Dfs root, the changes are automatically replicated to other

replica members.

Dfs and file replication support the following features:

• Mulitmaster replication replicates modified files and modified ACLs when a file is closed.
• Files can be modified on any replica member.
• Only Windows 2000 NTFS volumes has have the potential to replicate. Other shares can be

published as alternates, but no replication occurs.

• Replication is journal based.
• Replication is Remote Procedure Call (RPC) based.
• FRS topology follows Active Directory replication topology.

The process of Dfs replication consists of a number of steps:

• A file changes. This is noted when a user closes a file.
• NTFS makes an entry in the NTFS Change Log.
• FRS monitors the NTFS journal for changes to Dfs links.

=====================================================================

winser5.html PAGE 9 2002/01/05

• FRS makes an entry into its own journal.
• FRS generates a staging file of the file change.
• FRS holds on to changes until scheduled to replicate.
• The destination pulls the staging file and applies the new files.

Adding Replica Dfs Root Servers

Each DFs root or link can reference a replicated set of shared resources.

To add Dfs replica servers to a Dfs domain root or link, right-click the Dfs root in the Distributed

File System Manager too, click New, and then click Root Replica. Enter the UNC path for the

replica server and share.

Enabling Dfs Replication

Dfs replication is disabled by default. To enable replication, right-click the Dfs root or Dfs link

in the Distributed file System snap-in, and then select Replication Policy. Highlight every server in

the replica set that you want to participate in FRS replication.

Configuring FRS for Inter-Site Replication

You can configure inter-site replication by using the Active Directory Sites and Services snap-in.

To configure the FRS settings, you must create a new site link for the inter-site transport protocol

listed in the console tree. Once you’ve created the site link, right-click the site link object and click

Properties.

Lesson Summary:

• FRS is the automatic file replication service in Windows 2000 Server.
• There are two types of replication: intra-site replication and inter-site replication.
• Within a site, a process called KCC automatically generates a ring topology for replication

among domain controllers in the same domain.

• Implementing FRS consists of several phases, including replicating SYSVOL, replicating

domain DFs roots, and configuring FRS.