Chapter 9, Implementing Windows Internet Name Service

Chapter 9, Lesson 1

Introducing NetBIOS

1. Introduction

A. In 1993, when Microsoft first introduced networking into the Windows operating systems, it relied on the NetBIOS Extended User Interface (NetBEUI) protocol for all communications.

1. NetBEUI relies on the NetBIOS name space as the sole identifier of the computers on the network.

B. Later, Microsoft replaced NetBEUI with Transmission Control Protocol/Internet Protocol (TCP/IP) as the default Windows networking protocol but retained the NetBIOS name space.

1. The resulting implementation is more commonly called NetBIOS over TCP/IP (NetBT).

C. NetBIOS is an application programming interface (API) that enables user applications to submit network input/output (I/O) and control directives to underlying network protocols.

D. NetBIOS provides commands and support for the following services:

1. Network name registration and verification

2. Session establishment and termination

3. Reliable connection-oriented session data transfer

4. Unreliable connectionless datagram data transfer

5. Support protocol (driver) and adapter monitoring and management

2. NetBIOS Naming

A. The NetBIOS name space was originally designed to be the only means of identifying specific computers used on a network.

|1| 1. The NetBIOS name itself is a 16-byte character string assigned by a network administrator to a particular entity on a network.

2. The NetBIOS name is either a unique (exclusive) or group (nonexclusive) name.

B. NetBIOS resource identifiers

1. To identify the type of resource a NetBIOS name identifies, the sixteenth character of the name is reserved for a resource identifier code.

a. This leaves 15 characters for the NetBIOS name itself.

b. Names shorter than 15 characters are padded with spaces (after the name) to ensure that the resource identifier code always appears as the sixteenth character.

C. The NetBIOS name assigned to a computer must be unique.

1. The NetBIOS name space consists of only a single level; there are no separate domain and host names, as in a Domain Name System (DNS) name.

2. Because the NetBIOS name space is flat (as opposed to being hierarchical), only one computer on the network can use a particular name.

D. Windows 2000 uses DNS for name resolution by default; on a network consisting only of computers running Windows 2000, NetBIOS is not needed.

1. All Windows 2000 operating systems include NetBIOS client capabilities (including a WINS client).

2. All Windows 2000 Server operating systems include a WINS server, only for reasons of interoperability with computers running earlier versions of Windows.

3. Registering and Resolving Names

A. As used with NetBEUI, the NetBIOS name space is a suitable solution for a small local area network (LAN), but it cannot support large enterprise internetworks.

1. Using the NetBIOS name space with TCP/IP instead of NetBEUI provides the scalability inherent in the hierarchical addressing system of the Internet Protocol (IP).

2. The relationship between NetBIOS names and IP addresses is similar to the relationship between DNS names and IP addresses.

B. Two basic functions are involved in the use of NetBIOS names with TCP/IP: name registration and name resolution.

1. As with DNS, name resolution is the process by which a computer uses a name to discover the IP address associated with that name.

2. Name registration is the process that enables a computer to use a particular name, either by adding it to a lookup table or by ensuring that no other computer is using that name, or both.

|2| C. There are several NetBIOS name resolution mechanisms, and each one has its own form of name registration. These NetBIOS name resolution mechanisms are

1. NetBIOS name caching

2. Lmhosts files

3. Broadcast transmissions

4. WINS

5. DNS and Hosts files

|3| 4. NetBIOS Name Caching

A. DNS name servers maintain a cache in which they store information about recently resolved names.

1. With NetBIOS, individual computers cache name resolution data.

2. Every computer running Windows that uses NetBIOS maintains a NetBIOS name cache in memory that contains the names it has recently resolved.

3. Entries remain in the NetBIOS name cache for a relatively short period of time (10 minutes for Windows 2000, by default).

4. Because the cache is accessed from memory, it is the fastest NetBIOS name resolution mechanism by far.

B. No matter what other mechanism the computer is configured to use, it always checks the cache first to see if the requested name is present.

C. Name registration for the cache is, for the most part, automatic.

1. Resolving a NetBIOS name by any means causes the name and its IP address to be added to the cache for later use.

2. When the entry’s lifetime expires, it is automatically removed from the cache.

D. It is possible to manually load the NetBIOS name cache using specially tagged entries in an Lmhosts file.

1. When you preload the cache from an Lmhosts file, the entries you create do not have a limited life like the automatically created entries do.

2. Using this method can accelerate the name resolution process permanently for certain systems.

5. Using Lmhosts

A. The Lmhosts file is the NetBIOS equivalent of the Hosts file sometimes used by DNS clients (and used by all TCP/IP systems before the introduction of DNS).

1. As with Hosts, Lmhosts is a lookup table stored on a local drive that contains a list of names and their equivalent IP addresses.

2. Differences between Lmhosts and Hosts

a. Lmhosts is for NetBIOS names, while Hosts is for host names.

b. Lmhosts includes a number of optional tags that perform other functions; Hosts has no additional tags.

B. In the days before WINS, the main function of the Lmhosts file was to provide name resolution for computers on different LANs.

1. Because the broadcast name resolution method can resolve only the names of computers on the local network, an Lmhosts file containing the names and addresses of key servers on other networks was needed to resolve their names.

C. An advantage of Lmhosts is that the file is stored locally and requires no network communications of any kind to resolve a name.

D. The disadvantage of Lmhosts, as with Hosts, is that the name registration process is manual; you must create the individual entries in the file on every computer.

E. Implementing Lmhosts

1. The format of a standard Lmhosts entry is the same as the format of a Hosts entry: the IP address comes first, followed by the NetBIOS name, separated by at least one space.

|4| 2. In Windows 2000, use the following procedure to enable Lmhosts:

a. Click Start, point to Settings, and then select Network And Dial-Up Connections to open the Network And Dial-Up Connections window.

b. Select the Local Area Connection icon, and then select Properties from the File menu to open the Local Area Connection Properties dialog box.

c. In the General tab, select Internet Protocol (TCP/IP) in the components list, and then click Properties to open the Internet Protocol (TCP/IP) Properties dialog box.

d. Click Advanced to open the Advanced TCP/IP Settings dialog box.

e. Select the WINS tab to display the dialog box.

f. Select the Enable LMHOSTS Lookup check box, and then click OK three times to close the dialog boxes you opened.

3. Enabling the Lmhosts file causes Windows to use it to resolve NetBIOS names when the other name resolution mechanisms have failed.

F. Using Lmhosts tags

1. In addition to simple name and IP address listings, an Lmhosts file can also contain specially tagged entries that perform other functions, such as preloading the NetBIOS name cache.

2. In most cases, the tags appear after the IP address and NetBIOS name of a standard Lmhosts entry and are preceded by a pound (#) symbol.

|5| 3. Slide 5 shows a sample Lmhosts entry. The various Lmhosts tags and their functions are

a. #PRE. Part of a NetBIOS name–to–IP-address mapping entry that causes that entry to be preloaded into the NetBIOS name cache

b. #DOM:domain. Part of a NetBIOS name–to–IP address mapping entry that indicates that the IP address belongs to a domain controller in the domain specified by domain

c. #INCLUDE filename. Forces the system to seek the file specified by the filename variable and parse it as if it were local

d. #BEGIN_ALTERNATE. Used to group multiple #INCLUDE statements

(1) Any single successful #INCLUDE statement falling between a #BEGIN_ALTERNATE tag and an #END_ALTERNATE tag causes the group to succeed.

e. #END_ALTERNATE. Used to mark the end of an #INCLUDE statement grouping

f. #0xnn. Provides support for nonprinting characters in NetBIOS names

g. #MH. Part of a NetBIOS name–to–IP-address mapping entry that defines the entry as a unique name that can have more than one address

h. #SG:name. Part of a NetBIOS name–to–IP-address mapping entry that associates that entry with a user-defined special (Internet) group specified by name

6. Using Broadcast Transmissions

A. The original NetBIOS name resolution mechanism used by Windows involves the transmission of broadcast messages to the local network.

1. NetBEUI networks use this method to discover the hardware addresses associated with specific NetBIOS names.

2. Windows TCP/IP networks not running WINS use it to resolve NetBIOS names into IP addresses.

|6| B. Broadcast name resolution

1. The computer wanting to resolve a NetBIOS name generates a NAME QUERY REQUEST message containing the name to be resolved and broadcasts it in the local network.

2. The message is packaged in a User Datagram Protocol (UDP) packet and addressed to port number 137.

3. Every computer running Windows on the network that is configured to use NetBIOS receives the broadcast transmissions and reads the NetBIOS name enclosed within them.

4. If a computer detects its own NetBIOS name in the message, it replies to the sender with a POSITIVE NAME QUERY RESPONSE message containing its IP address.

5. Computers that do not recognize the NetBIOS name in the NAME QUERY REQUEST message discard it.

|7| C. Broadcast name registration

1. Name registration ensures that no two computers on the same network are using the same NetBIOS name.

2. The process begins with the computer generating a NAME REGISTRATION REQUEST message containing its own NetBIOS name and broadcasting it over the network three times.

3. The computers receiving the broadcasts process them in the same way that they do NAME QUERY REQUESTS.

4. If a computer recognizes its own NetBIOS name in a NAME REGISTRATION REQUEST message, it responds by transmitting a NEGATIVE NAME REGISTRATION RESPONSE to the original computer as a unicast.

a. This message denies the name registration and causes the computer to prompt the user for a different NetBIOS name.

5. When the computer trying to register its name receives no responses to its broadcasts, it generates a NAME OVERWRITE DEMAND message and broadcasts it over the network.

a. In effect, the computer declares that it is taking possession of the name.

D. The broadcast transmission method is a simple but effective name resolution mechanism that is well suited to small network installations.

E. Disadvantages of the broadcast transmission method

1. The primary disadvantage is that broadcast transmissions are limited to the local network, making it impossible for a computer to resolve the names of computers on other networks using this method.

2. Another significant drawback is the amount of resources used.

a. Broadcasts are inherently inefficient, and NetBT broadcasts are particularly so.

b. Every NAME REGISTRATION REQUEST and NAME QUERY REQUEST transmitted by a NetBIOS client must be processed by every computer on the network.

c. The data-link layer protocol on each computer receives the packet and passes it up through the protocol stack all the way to the NetBIOS interface, where the system then decides whether to respond to the message or discard it.

d. The majority of the broadcast messages are discarded by each computer, meaning that the bandwidth and CPU time used to process them is wasted.

7. Understanding NetBIOS Node Types

A. The NetBT standards define a series of node types, each of which consists of a number of NetBIOS name resolution mechanisms that the computer uses in a particular order.

1. Each mechanism acts as a fallback for its predecessor, providing the system with every opportunity to resolve a specific name.

|8| B. The NetBT node types

1. B node (broadcast node). Calls for the exclusive use of the broadcast method for NetBIOS name registration and resolution

2. P node (point-to-point node). Calls for the exclusive use of NetBIOS name servers (that is, WINS servers) for NetBIOS name registration and resolution

3. M node (mixed mode node). Calls for the exclusive use of the broadcast method for name registration

a. For NetBIOS name resolution, the m node uses broadcasts first.

b. If the broadcast method fails to resolve a name, the m node uses a NetBIOS name server.

C. Microsoft node types

1. When Microsoft designed WINS to function as a NetBIOS name server, the three NetBT node types were inadequate.

a. The b node and p node types had no fallbacks.

b. The m node type used a NetBIOS name server (NBNS) only for resolving names of computers on other networks.

2. To address these shortcomings, Microsoft developed the following additional node types:

|9| a. Modified b node. Calls for the exclusive use of the broadcast method for name registration

(1) For name resolution, this type of node uses broadcasts first and the Lmhosts file next if broadcasts fail to resolve a name.

(2) This is the default node type for a computer running Windows that is not configured to use WINS.

b. H node (hybrid node). Calls for the exclusive use of NetBIOS name servers for name registration

(1) For name resolution, this type of node uses NetBIOS name servers first and the broadcast method next if the NetBIOS name servers fail to resolve the name or are unavailable.

c. Microsoft-enhanced h node. Windows includes options that can supplement an h node system with alternatives such as Lmhosts name resolution, Windows Sockets calls to a DNS server, and a Hosts file.

(1) All of the alternatives are used if both WINS servers and broadcasts fail to resolve a name.

Chapter 9, Lesson 2

WINS Name Registration and Resolution

1. Introduction

A. Although Windows 2000 does not use NetBIOS by default, Windows 2000 Server still includes WINS to support clients running earlier versions of Windows.

B. WINS is a NetBIOS name server that provides name registration and resolution services for NetBIOS clients anywhere on a network.

1. Using WINS prevents NetBIOS clients from having to use broadcast transmissions and eliminates the need for administrators to maintain Lmhosts files.

2. Understanding WINS Messaging

A. WINS uses the same message formats as broadcast name resolution, but all the transmissions are unicasts.

1. Only the WINS client and server expend processing resources during the name registration and resolution processes.

2. Unicast transmissions are not limited by network boundaries, meaning that one WINS server can service clients all over an internetwork.

3. A WINS client can communicate with WINS servers using unicast transmissions because it is already configured with the IP addresses of one or more WINS servers as part of its TCP/IP configuration.

4. The result is that WINS is the most efficient NetBIOS name resolution mechanism by far.

B. WINS name registration

1. On a WINS network, every client configured with WINS server addresses performs a name registration procedure as the system starts.

2. This procedure adds the client’s NetBIOS name and IP address to the WINS server’s database so that other computers on the network can resolve the name.

3. The WINS server checks the NetBIOS name against its database to ensure that no two computers on the network are using the same NetBIOS name.

4. The WINS name registration process is initiated by the client, which transmits a NAME REGISTRATION REQUEST message to the first address in its list of WINS servers.

a. The client sends this message as a unicast in a UDP packet addressed to port number 137.

(1) If the WINS server is available to process the message, it checks to see if another computer has already registered the client’s NetBIOS name.

(2) If the name does not exist in the WINS database, the server adds a NetBIOS name–to–IP-address mapping to its database and returns a POSITIVE NAME REGISTRATION RESPONSE message to the client, also as a unicast.

b. This response message includes a Time to Live (TTL) value, which is the amount of time that the NetBIOS name will stay registered to the client without being renewed.

5. When a WINS client attempts to register a name that is already in the WINS database, the WINS server performs a name challenge process to ensure that the name is actually still in use.

a. The name challenge begins with the WINS server transmitting a series of NAME QUERY REQUEST messages to the currently registered owner of the NetBIOS name.

(1) If the name’s current owner responds to the server with a POSITIVE NAME QUERY RESPONSE message, the server sends a NEGATIVE NAME REGISTRATION RESPONSE message to the new applicant.

(a) This message denies the name registration and forces the user at the client computer to choose a new NetBIOS name.

(2) The server purges the name from its database and assigns it to the new client if one of the following happens:

(a) The WINS server does not receive a response after transmitting three NAME QUERY REQUEST messages at 500-millisecond intervals.

(b) The currently registered name owner responds with a NEGATIVE NAME QUERY RESPONSE message, indicating that it is no longer using the name.

C. WINS name renewal

1. To prevent leaving obsolete information in the database, WINS servers register all NetBIOS names for a specified interval, the TTL, which is six days by default.

2. Each time a WINS client computer restarts and registers its name with the WINS server, the TTL interval is reset.

3. If the client remains logged on to the network continuously for half of the TTL interval (three days, by default), it begins transmitting NAME REFRESH REQUEST messages to the WINS server.

a. The server then replies with one of the following messages:

(1) A POSITIVE NAME REFRESH RESPONSE message, which resets the TTL timer

(2) A NEGATIVE NAME REFRESH RESPONSE message, which cancels the name registration and forces the client to register a different NetBIOS name

b. If the client receives no response from the server, it repeats its request every two minutes until half of the remaining TTL interval has passed.

c. The client then switches to the second WINS server specified in its configuration and transmits the same NAME REFRESH REQUEST messages.

d. If the client receives no response, it retransmits every two minutes until half of the remaining TTL interval has expired.

e. The renewal process continues in this way, switching WINS servers every time the client reaches half of the remaining TTL value.

f. If the TTL period expires without the client receiving a response from a WINS server, the client reverts to broadcast name registration.

D. WINS name release

1. As part of a WINS client’s system shutdown sequence, the computer transmits a NAME RELEASE REQUEST message to the WINS server, indicating that it is no longer using its registered NetBIOS name.

2. The server responds to the request with one of the following messages:

a. A POSITIVE NAME RELEASE RESPONSE message, indicating that the server has successfully released the name

b. A NEGATIVE NAME RELEASE RESPONSE message, which only occurs when the record for the NetBIOS name contains a different IP address than that of the computer sending the message

E. WINS name resolution

1. When a WINS client initiates the process of communicating with another NetBIOS computer, it first checks its NetBIOS name cache to see if there is an entry for the desired name.

2. If the client cannot resolve the name from its cache, it transmits a NAME QUERY REQUEST message to its primary WINS server.

3. The WINS server searches its database for a NetBIOS
name–to–IP-address mapping for the name specified in the query and returns the IP address to the WINS client in a POSITIVE NAME QUERY RESPONSE message.

4. If the requested name does not exist in the WINS database, the server transmits a NEGATIVE NAME QUERY RESPONSE back to the client.

5. If the WINS server experiences a delay in responding to the client’s request, it might send a series of WAIT FOR ACKNOWLEDGMENT RESPONSE (WACK) messages to prevent the client from timing out.

6. If the client receives no response at all from the WINS server after a given amount of time, it switches to the next WINS server address that it is configured to use (if there is one) and transmits another series of NAME QUERY REQUEST messages.

7. This process is repeated until the client has sent messages to all the configured WINS servers (up to 12, in the case of Windows 2000).

8. If there is still no response, or if any server transmits a NEGATIVE NAME QUERY RESPONSE message, the client switches to an alternate NetBIOS name resolution mechanism, such as broadcast query messages, as defined by its node type.

9. If there is no alternative name resolution method, the name resolution process fails.

3. Understanding the NetBT Message Formats

A. The NetBT messages used in WINS communications are derived from the DNS packet format, with additional types and codes added to support NetBIOS functions.

1. As with DNS, the message consists of a header and four sections called Question, Answer, Authority, and Additional.

2. WINS uses only the header and the Question and Answer sections.

|10| B. The Header section

1. The format of the NetBT message header is shown on Slide 10.

2. Functions of the NetBT header fields

a. NAME_TRN_ID (2 bytes). Contains a transaction identifier that computers use to match request messages with responses

b. OPCODE (5 bits). The first bit of the OPCODE field is the R flag, which specifies whether the message is a request or a response.

(1) The remaining four bits specify the type of message contained in the packet, using values representing the following types: query, registration, release, WACK, and refresh.

c. NM_FLAGS (7 bits). Contains seven 1-bit flags that provide information about the message and the server that generated it

d. RCODE (4 bits). Used in response packets to specify the results of a particular request

e. QDCOUNT (2 bytes). Specifies the number of entries in the Question section of the message

(1) NetBT messages always contain only one entry in the Question section.

f. ANCOUNT (2 bytes). Specifies the number of resource records in the Answer section of the message

(1) NetBT messages always contain only one entry in the Answer section.

g. NSCOUNT (2 bytes). Specifies the number of resource records in the Authority section of the message

(1) The Authority section is not used in NetBT messages.

h. NRCOUNT (2 bytes). Specifies the number of resource records in the Additional section of the message

(1) The Additional section is not used in NetBT messages.

C. The Question section

1. Typically contains a NetBIOS name to be registered or resolved

2. Only the following message types contain information in the Question section:

a. NAME REGISTRATION REQUEST

b. NAME REFRESH REQUEST

c. NAME RELEASE REQUEST

d. NAME OVERWRITE DEMAND

e. NAME QUERY REQUEST

3. When the Question section contains information, the value of the QDCOUNT field in the header is set to 1, and three additional fields appear immediately after the header.

|11| 4. The format of the Question section is shown on Slide 11. The functions of the fields are

a. QUESTION_NAME (variable). Contains the NetBIOS name that the message is intended to register or resolve

b. QUESTION_TYPE (2 bytes). Contains a code that specifies the type of request in the message

c. QUESTION_CLASS (2 bytes). Contains a code that specifies the class of the request, for which only one possible value exists: 0x0001, indicating the Internet class

D. The Answer section

1. As with DNS, a NetBT message contains three sections that can carry the NetBIOS equivalent of a DNS resource record: the Answer, Authority, and Additional sections.

a. NetBT messages on a Windows network never need to carry more than one resource record, so they only use the Answer section.

b. The maximum possible value of the header’s ANCOUNT field is 1.

2. The only NetBT messages that contain information in the Answer section are the positive response and WACK messages, as follows:

a. POSITIVE NAME REGISTRATION RESPONSE

b. POSITIVE NAME QUERY RESPONSE

c. POSITIVE NAME REFRESH RESPONSE

d. POSITIVE NAME RELEASE RESPONSE

e. WAIT FOR ACKNOWLEDGMENT RESPONSE

|12| 3. A NetBT message containing an Answer section includes several resource record fields, with the following functions:

a. RR_NAME (variable). Contains the NetBIOS name from the QUESTION_NAME field in the request message to which this is the response

b. RR_TYPE (2 bytes). Contains a code that specifies the type of resource record in the Answer section

c. RR_CLASS (2 bytes). Contains a code that specifies the class of the resource record in the Answer section

d. TTL (4 bytes). Specifies the Time to Live value for the information in the resource record

e. RDLENGTH (2 bytes). Specifies the number of bytes in the RDATA field

f. RDATA (variable). Contains the IP address of the system with which the NetBIOS name in the RR_NAME field is associated

(1) Also includes bits that indicate whether the name is a unique or group name and specify the node type of the computer

Chapter 9, Lesson 3

Implementing WINS

1. Installing a WINS Server

A. WINS is supplied with Windows 2000 Server, but it is not installed by default.

1. You can install WINS with the operating system, or you can install it separately later, using Add/Remove Programs on Control Panel.

2. The server must be configured with a static IP address, subnet mask, and default gateway.

a. Do not use DHCP to configure the TCP/IP configuration parameters for a WINS server.

|13| B. To install the WINS service:

1. On a computer running Windows 2000 Server, log on as Administrator.

2. Click Start, point to Settings, and then click Control Panel.

3. Double-click the Add/Remove Programs icon, and then click Add/Remove Windows Components to activate the Windows Components Wizard.

4. In the Components list, scroll down to select Networking Services.

5. Click Details to open the Networking Services dialog box.

6. In the Subcomponents Of Networking Services list, select the Windows Internet Name Service (WINS) check box.

7. Click OK, and then click Next.

a. If prompted, type the full path to the Windows 2000 distribution files and then click Continue. Required files are copied to your hard disk.

8. Click Finish to close the Windows Components Wizard.

C. After you install WINS on the computer running Windows 2000 Server, you should configure the server to function as a WINS client, with itself as the WINS server.

2. Supporting Non-WINS Clients

A. In a WINS environment, you can provide support for non-WINS clients in either of two ways: by using static mappings or by configuring a WINS proxy agent.

B. Using static mappings

1. On a network that includes NetBIOS computers that are not WINS clients, you can configure a static NetBIOS name–to–IP-address mapping for each non-WINS client.

a. If you have DHCP clients that require static mappings, you must create IP address reservations for them on the DHCP server so that their IP addresses are always the same.

2. To configure a static mapping for a non-WINS client:

a. Click Start and from the Administrative Tools program group, open the WINS console.

b. Expand the icon for your WINS server, and then select Active Registrations.

|14| c. From the Action menu, select New Static Mapping to display the New Static Mapping dialog box.

d. In the Computer Name text box, type the NetBIOS name of the computer you want to create a mapping for.

e. In the Type drop-down list, select the type of mapping you want to create.

f. In the IP Address text box, type the IP address that you want to map to the NetBIOS name you selected.

g. Click OK to create the static mapping.

C. Configuring a WINS proxy agent

|15| 1. A WINS proxy agent extends the name resolution capabilities of the WINS server to non-WINS clients by listening for broadcast name registration and name resolution requests and then forwarding them to a WINS server.

a. When a non-WINS client broadcasts a NAME REGISTRATION REQUEST message, the WINS proxy agent forwards the request to the WINS server to verify that no other WINS client has registered that name.

b. When a WINS proxy agent detects a NAME QUERY REQUEST broadcast, it checks its NetBIOS name cache and attempts to resolve the name.

(1) If the name is not in the cache, the proxy agent sends the request to a WINS server.

c. The WINS server then responds to the WINS proxy agent with the IP address for the requested NetBIOS name.

(1) The WINS proxy agent returns this information to the non-WINS client.

2. To configure a WINS proxy agent, edit the registry on a WINS client computer by setting the value for the EnableProxy entry to 1, and then restart the computer.

a. The EnableProxy entry is located in the registry under the subkey HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\
NetBT\Parameters.

3. Maintaining the WINS Database

A. The WINS server included with Windows 2000 stores information about the computers on the network in a database.

1. Adding and removing records can leave empty spaces that waste storage space.

a. You should periodically compact the WINS database using the Jetpack.exe utility included with Windows 2000.

|16| B. To manually compact the WINS database:

1. Open a Command Prompt window on the WINS server.

2. Change to the systemroot\System32\Wins folder on the system drive.

3. Stop the WINS Server service by typing net stop wins at the command prompt.

4. Compact the database by typing jetpack wins.mdb tmp.mdb at the command prompt.

5. Restart the WINS Server service by typing net start wins at the command prompt.

6. Close the Command Prompt window.

Chapter 9, Lesson 4

Configuring WINS Replication

1. Replication Overview

A. WINS database replication occurs whenever the database changes, including when a NetBIOS name is released.

1. Replicating databases enables a WINS server to resolve the NetBIOS names of hosts registered with another WINS server.

B. To replicate database entries, each WINS server must be configured as either a pull or a push partner with at least one other WINS server.

1. A push partner is a WINS server that sends a message to its pull partners notifying them when its WINS database has changed.

a. When a WINS server’s pull partners respond to the message with a replication request, the WINS server sends (pushes) a copy of its new database entries to its pull partners.

2. A pull partner is a WINS server that requests new database entries from its push partners by requesting entries with a higher version number than the entries it received during the last replication.

2. Configuring a WINS Server as a Push or Pull Partner

|17| A. Rules for configuring WINS server replication

1. Configure a push partner when servers are connected by fast links, because push replication occurs when a specified number of updated WINS database entries is reached.

2. Configure a pull partner between sites, especially across slow links, because pull replication can be configured to occur at specific intervals.

3. Configure each server to be both a push partner and a pull partner to replicate database entries between them.

4. Every WINS server must be both a push partner and a pull partner for the replication to be complete, but not necessarily partners with each other.

|18| B. In the example shown on Slide 18, all WINS servers at both the Sydney and Seattle sites push their new database entries to a single server at their local site.

1. The server at each site that receives the push replication is configured for pull replication with its counterpart at the other site.

a. The server is configured in this way because the network link between Sydney and Seattle is relatively slow.

b. You can configure pull replication to occur when the link is used the least.

2. The single server at each site is also configured to push the database entries it receives from the other site to all the local WINS servers.

a. This ensures that all the WINS servers on the network are updated regularly.

|19| C. Ring replication topology

1. Each WINS server is configured to be a push partner of its clockwise neighbor in the ring and a pull partner of its counterclockwise neighbor.

a. Although each WINS server must be both a push and a pull partner, you do not have to pair the servers and make them push and pull partners of each other.

2. One of the potential problems with ring replication topology on a network with wide area network (WAN) links is that the failure of a single WAN link can prevent all the WINS servers from being properly replicated.

|20| a. One solution to this problem is to create two counter-rotating rings using different links.

b. This topology provides greater fault tolerance because multiple link failures would have to occur for the replication process to be interrupted.

3. Planning How Many WINS Servers to Use

A. On a smaller network, a single WINS server can adequately service up to 10,000 clients for NetBIOS name resolution requests.

1. To provide fault tolerance, you can configure a second computer to function as a backup WINS server.

2. If you use only two WINS servers, you can easily set them up as replication partners of each other.

3. For simple replication between two servers, one server should be set as a pull partner and the other as a push partner.

B. A larger network sometimes requires more WINS servers for several reasons including, most importantly, the number of client connections per server.

1. The number of users that each WINS server can support varies with usage patterns, data storage, and the processing capabilities of the WINS server computer.

2. Some enterprise network environments require more robust hardware to handle WINS activity, so you might benefit from upgrading the server computer.

3. When planning your servers, remember that each WINS server can simultaneously handle hundreds of registrations and queries per second.

4. Initiating Database Replication

A. Database replication requires that you configure at least one push partner and one pull partner.

|21| B. There are four conditions that initiate the replication of the WINS database:

1. At system startup

2. At a configured interval, such as every five hours

3. When a WINS server has reached a configured threshold for the number of registrations and modifications to the WINS database

4. By forcing replication in the WINS console

5. WINS Automatic Replication Partners

A. If your network supports multicasting, you can configure the WINS server to automatically find other WINS servers on the network by multicasting to the IP address 224.0.1.24.

1. This multicasting occurs by default every 40 minutes.

2. Any WINS servers found on the network are automatically configured as push and pull replication partners, with pull replication set to occur every two hours.

3. If network routers do not support multicasting, the WINS server finds only other WINS servers on its subnet.

4. Automatic WINS server partnerships are turned off by default.

6. Configuring WINS Database Replication

A. To configure a WINS server to perform database replication with another WINS server:

1. Log on as Administrator to the computer running WINS.

2. Click Start and from the Administrative Tools program group, open the WINS console.

3. Right-click the Replication Partners folder under your WINS server, and then from the pop-up menu, select New Replication Partner to display the New Replication Partner dialog box.

4. Type the NetBIOS name or IP address of another WINS server on the network in the WINS Server text box, or click Browse to select a server, and then click OK.

a. The Replication Partners list appears in the detail pane of the console with the server you added to the list.

5. Right-click the replication partner that you just added in the detail pane, and from the pop-up menu, select Properties to display the Server Properties dialog box.

6. Select the Advanced tab.

7. In the Replication Partner Type drop-down list box, select Push/Pull.

a. The pull replication interval is set for 30 minutes by default.

8. In the Push Replication box, set the value in the Number Of Changes In Version ID Before Replication box to 5.

a. You can modify the push and pull replication intervals as needed.

9. Click OK to close the Properties dialog box.

10. Repeat the entire process (steps 1–8) for the partner WINS server on your network.

B. After you have configured your WINS servers to be replication partners, you can force the replication process to begin.

1. To do this, right-click the Replication Partners folder for one of your servers and, from the pop-up menu, select Replicate Now.