Deploying Medium and Large Server Farms

Archived content. No warranty is made as to technical accuracy. Content may contain URLs that were valid when originally published, but now link to sites or pages that no longer exist.

Published: June 9, 2004 | Updated : March 1, 2006

This is a sample chapter from the Microsoft SharePoint Products and Technologies Resource Kit. You can obtain the complete resource kit (ISBN 0-7356-1881-X), which includes a companion CD-ROM, from Microsoft Press.

This chapter discusses how to deploy a medium or large server farm. The majority of the chapter consists of step-by-step instructions that can be used as a guide for deploying a farm. This chapter does not cover how to determine which type of farm to use in given situations. For that type of information, please refer to Chapter 9, “Capacity Planning.”

This chapter should be viewed as a getting-started guide for building a medium or large server farm. It will cover everything necessary for building both a medium server farm and a larger server farm. It will also contain instructions for how to migrate from a medium farm to a large farm.

On This Page

Topologies
Summary

Topologies

Many permutations of topologies can be built using Microsoft Office SharePoint Portal Server 2003, but only a small number of these topologies are supported by Microsoft. The simplest topologies are the single-server and small-server-farm topologies, which have been covered in previous chapters.

Medium and large server-farm topologies can get much more complex. A medium server topology requires at a minimum one front-end Web server running the search application, one index/job server running SharePoint Portal Server 2003, and one database server running Microsoft SQL Server 2000 with Service Pack 3a or later. Additional servers can be added to this topology to enable higher availability, higher capacity, or both, but you cannot have a medium farm with fewer than four servers.

Several topologies are not supported by Microsoft. When you install one of these topologies, it will appear to function—that is, you can create the topology on the Components Selection page, but it will not be eligible for support and you will not be able to create a new portal site in it. In addition, unsupported topologies cannot be backed up or restored. When you open SPSBackup.exe, it will show an error “Topology not supported” and exit. It is strongly recommended that you stay away from these unsupported topologies. Installing the database component on one of the servers running SharePoint Portal Server is an unsupported topology for a farm. Installing two or more servers with all (Web/search/index) components is also not supported.

Here are the supported topologies:

Small Server Farm.

• One server running SQL Server 2000 and one server running SharePoint Portal Server 2003 assigned the Web, Search, Job, and Index services.

Medium Server Farm.

• One or two servers running SharePoint Portal Server 2003 assigned the Web service (more commonly known as front-end Web servers) and running SharePoint Portal Server 2003 assigned the Search, Job, and Index services; and one server running SQL Server 2000.

Large Server Farm.

• Two to eight servers running SharePoint Portal Server 2003 assigned the Web service (more commonly known as front-end Web servers), two to four servers running SharePoint Portal Server 2003 assigned the Search service, one to four servers running SharePoint Portal Server 2003 assigned the Index service (one of which must be assigned the Job Server role), and any number of servers running SQL Server 2000.

In all supported topologies, you can use a single additional server running SharePoint Portal Server 2003 for the purpose of running the backward-compatible document library.

Two factors differentiate a medium farm from a large farm. The first factor is the server component matrix—you will need more servers to run a minimum large farm. The second difference is that the Web and search components run on all front-end Web servers in a medium farm, while on a large farm these components must each run on separate servers. Medium server farms become large server farms by adding Web/search front-end servers, using Network Load Balancing (NLB) to distribute the load, and by adding search and indexing servers to the server component matrix. As long as you stay within the boundaries of a prescribed server farm, any combination of these can be employed. Only four index servers can exist in a large server farm because each search server can only consume four separate catalogs. For more information about how each of these components affects performance, please see Chapter 9. The most common medium-farm topology consists of two front-end Web servers running search and indexing/job servers, and one SQL component, for a total of four servers. The SQL component can be either a single server or a clustered server, but because the cluster acts like one system, it will be discussed here as one component. SharePoint Portal Server does not really care what the SQL Server topology is because this is abstracted from the server farm deployment. Therefore, you can be running any supported SQL topology and SharePoint Portal Server views it as a single-server installation for topology purposes. This is the topology that will be covered when discussing medium server farms in this chapter. Because this topology employs NLB as well as all other components required by a medium farm, the instructions provided can be used to easily add servers to the topology.

As mentioned previously, a large server farm is very similar to a medium server farm, except that the search and Web components run on separate systems and large server farms can support more servers than a medium farm. The minimum topology for a large server farm is two front-end Web servers running SharePoint Portal Server 2003, two search servers running SharePoint Portal Server 2003, one index management server running SharePoint Portal Server 2003, and one database component running Microsoft SQL Server 2000 Service Pack (SP) 3a. The SQL component can be either a single server or a clustered server, but because the cluster acts like one system, it will be discussed here as one component. The large server farm can be scaled out by adding front-end Web servers, search servers, or index management servers. You cannot have more than four search servers or four index management servers in any farm. Because only four catalogs can be consumed by the search servers, having more than four index management servers will cause search to function improperly, as some content sources will not be searched.

Both medium and large server farms support running one server with the backward- compatible document libraries. There can only be one server in the farm that has these components. In addition, each medium and large server farm requires one job server. The job server role must be assigned to an index management server, in part, because the job server is responsible for indexing all portal site content and people. Here is a complete list of all the job server activities:

• Managing the indexing of all portal content

• Indexing people from the profile database

• Hosting the Single Sign-On administration pages

• Performing Audience calculations

• Running the Alert service for the server farm

• Importing profiles from Active Directory

The job server is an important role, but one that can be assigned only to an individual server in the server farm. This is an example of a shared service—a service that runs on a single server in the server farm but is consumed by all servers in the farm. Shared services is discussed more fully in Chapter 14, “Shared Services.”

The way that the servers know which role they have been assigned in the farm is handled by the Admin service. When you make a component change between servers in a server farm on the Component Selections page, this information is written to the Configuration database. The Admin service, which runs on each server running SharePoint Portal Server 2003, checks this configuration database for component assignment changes every 30 seconds. If it detects a change, the service turns on or off those portions of the SharePoint Portal Server code to run only the assigned services. For example, let’s assume you have a server in a large server farm that has been assigned the index and job server roles. Now, let’s assume that you decide to move the job server role to another index server in the farm. On both of these servers, all the SharePoint Portal Server code is installed, but when you make the service assignment change, the Admin service on the first server will “turn off” the job server role services and the Admin service on the second server will “turn on” the job server role services and assume these responsibilities. In this manner, each server in the farm will know its role in the farm, the services it is to run to “complete” the farm, and when to change its role to meet new role assignments you select on the Components Selection page.

Preparing for Deploying a Farm

Before building a server farm, you need to collect some information and create prerequisite accounts. Table 12-1 shows the information needed to create a portal site and includes the values for each item used for a medium farm in this chapter. The numbers and DNS entries filled in are examples. You should fill in your own IP Address Range and DNS entries based on your environment.

Table 12-1. Roles, IP Addresses, and Names in a Medium Farm

Physical Server	DNS Name	IP Address	Subnet Mask	Description	Notes
n/a	cluster.sps.test.local	172.16.12.200	255.255.0.0	Corporate portal virtual host	Used for NLB
SPS-01	SPS-01.sps.test.local	172.16.12.10	255.255.0.0	Host address for front-end Web/search server SPS-01	If only one NIC in server, bind this address and cluster address (below). If multiple NICs, choose the NIC connected to the internal network for this address.
SPS-01		172.16.12.200	255.255.0.0	Cluster address for front-end Web/search server SPS-01	If only one NIC in server, bind this address and host address (above). If multiple NICs, choose the NIC connected to the external network for this address.
SPS-02	SPS-02.sps.test.local	172.16.12.11	255.255.0.0	Host address for front-end Web server SPS-02	If only one NIC in server, bind this address and cluster address (below). If multiple NICs, choose the NIC connected to the internal network for this address.
SPS-02		172.16.12.200	255.255.0.0	Cluster address for front-end Web server SPS-02	If only one NIC in server, bind this address and host address (above). If multiple NICs, choose the NIC connected to the external network for this address.
SPS-03	SPS-03.sps.test.local	172.16.12.12	255.255.0.0	Index management server
SQL-01	SQL-01.sps.test.local	172.16.12.9	255.255.0.0	Address of SQL Server (or active server if cluster)	If clustered, this information should be used for the SQL cluster’s active node.
DocLib	DocLib.sps.test.local	172.16.12.15	255.255.0.0	Optional server for backward-compatible document libraries

For a large server farm, the information in Table 12-2 will be used.

Table 12-2. IP Addresses and Names in a Large Farm

Physical Server	DNS Name	IP Address	Subnet Mask	Description	Notes
n/a	cluster.sps.test.local	172.16.12.200	255.255.0.0	Corporate portal virtual host	Used for NLB
SPS-01	SPS-01.sps.test.local	172.16.12.10	255.255.0.0	Host address for front-end Web server SPS-01	If only one NIC in server, bind this address and cluster address (below). If multiple NICs, choose the NIC connected to the internal network for this address.
SPS-01		172.16.12.200	255.255.0.0	Cluster address for front-end Web server SPS-01	If only one NIC in server, bind this address and host address (above). If multiple NICs, choose the NIC connected to the external network for this address.
SPS-02	SPS-02.sps.test.local	172.16.12.11	255.255.0.0	Host address for front-end Web server SPS-02	If only one NIC in server, bind this address and cluster address (below). If multiple NICs, choose the NIC connected to the internal network for this address.
SPS-02		172.16.12.200	255.255.0.0	Cluster address for front-end Web server SPS-02	If only one NIC in server, bind this address and host address (above). If multiple NICs, choose the NIC connected to the external network for this address.
SPS-03	SPS-03.sps.test.local	172.16.12.12	255.255.0.0	Index management server
SPS-04	SPS-04.sps.test.local	172.16.12.13	255.255.0.0	Search server
SPS-05	SPS-05.sps.test.local	172.16.12.14	255.255.0.0	Search server
SQL-01	SQL-01.sps.test.local	172.16.12.9	255.255.0.0	Address of SQL Server (or active server if cluster)	If clustered, this information should be used for the SQL cluster’s active node.
DocLib	DocLib.sps.test.local	172.16.12.15	255.255.0.0	Optional server for backward-compatible document libraries

SharePoint Portal Server farms use many different service accounts. These service account settings can all use the same account, or they can use different accounts if corporate policies dictate. For the purposes of this chapter, all service account settings will use the same account. Table 12-3 shows the different service accounts.

Table 12-3. Service Accounts

Server	Service Name	Access Account	Role
SQL Server	MSSQLSERVER	domain\spssql	Database system administrator
SQL Server	SQLSERVERAGENT	domain\spssql	Database system administrator
Web front-end(s), search, and index	SharePointPS search	domain\spssql	SharePoint Portal Server administrator
Web front-end(s), search, and index	SharePoint administration	domain\spssql	SharePoint Portal Server administrator
Web front-end(s), search, and index	SharePoint portal alert	domain\spssql	SharePoint Portal Server administrator
Web front-end(s), search, and index	SharePoint timer service	domain\spssql	SharePoint Portal Server administrator
Web front-end(s), search, and index	Configuration database	domain\spssql	Administrator right on search and index servers
Web front-ends	Default content access account	domain\spssql	Used for crawling content on the Internet. This should have elevated privileges to access content throughout the intranet. Keep in mind that even though this account can access all data, users will see only results for content that they have access to.

The SharePoint Portal Server administrator must have at least local Power User rights on each server running SharePoint Portal Server. To grant the account Power User permissions, follow these steps:

1. Open Administrative Tools, click Computer Management.

2. Expand Local Users and Groups and click the Groups folder.

3. Double-click the Power Users group.

4. Click the Add button.

5. Enter the name of the service account (SPS\SPSSQL) and click OK. Click OK again to finalize the group membership addition.

Configuring Network Load Balancing

Network Load Balancing (NLB) is a clustering technology that allows multiple servers that are configured the same to service clients. It distributes the incoming connections between all the servers in the cluster. In a SharePoint Portal Server farm environment, NLB is used to distribute load between all the front-end Web servers in the farm. This gives the farm redundancy and the ability to scale out.

  Windows Network Load Balancing distributes load and provides failover for the Web servers only. Although the search services for the medium server farm operate on the Web server computers, search services use their own load-balancing mechanisms, and therefore Windows NLB does not affect them.

SharePoint Portal Server does not depend on NLB, so any load-balancing technology can be used. NLB is discussed here because it comes with all Windows Server 2003 editions. If your corporate environment uses a different load-balancing mechanism, feel free to use that instead.

NLB works by setting the same physical (MAC) address on all nodes of the cluster. In essence, this makes all the machines’ network cards act like a single network card. The MAC address used for NLB cannot be manually configured. It is set by Windows based on the IP address chosen for the cluster. The following sections detail configuring and testing NLB for use in a SharePoint Portal Server farm.

To configure the first front-end NIC (SPS-01)

1. In Control Panel, click Network Connections.

2. Right-click Local Area Network Connection and then click Properties.

3. Under This connection uses the following items, click Internet Protocol (TCP/IP).

4. Click Properties.

5. Click Use the following IP address, and type in the IP address with your assigned front- end static IP address. (This is the IP address for the machine, not the cluster.)

6. Type in the subnet mask associated with this IP address.

7. Type in the default gateway with your assigned front-end gateway.

8. Click Advanced.

9. Under IP address, click Add.

10. In the IP Address box, type the cluster IP address, as illustrated in Figure 12-1.

    

    Figure 12-1. Setting the cluster IP address on SPS-01

11. Type in the subnet mask associated with the cluster IP.

12. On the DNS tab, remove all DNS entries.

13. Deselect the Register this connections addresses in DNS check box.

14. On the WINS tab, remove all WINS entries.

15. On the LAN Connection Properties page, under This connection uses the following items, click Network Load Balancing.

16. Click Network Load Balancing again, and then click Properties.

17. On the Cluster Parameters tab, perform the following steps, as illustrated in Figure 12-2:

    1. In the IP Address box, type the cluster IP (172.16.12.200).

    2. Type the subnet mask associated with the cluster IP.

    3. In the Full Internet Name box, type the DNS name.

    4. Change the Cluster operation to the Unicast setting.

         The Unicast setting can be used only when your server has multiple Network Interface Cards (NIC), while Multicast will be used when there is only one NIC. For a further explanation, please see the Windows Server 2003 Resource Kit or the “Network Load Balancing: Configuration Best Practices for Windows 2000 and Windows Server 2003” white paper.

    5. Leave the Allow remote control check box unselected.

       

       Figure 12-2. Configuring NLB cluster properties on SPS-01

18. On the Host Parameters tab, perform the following steps, as illustrated by Figure 12-3:

    1. Because this is the first Web/search server, enter Priority 1.

    2. In the Dedicated IP Configuration section, enter the address and subnet mask that you assigned to this machine in step 5.

    3. Do not change the initial host state settings.

       

       Figure 12-3. Setting NLB host parameters on SPS-01

19. On the Port Rules tab, in the Filtering mode section, verify that multiple host is enabled and that Portal Server is set to None, as shown in Figure 12-4.

    

    Figure 12-4. Set NLB Port Rules on SPS-01

To configure additional front-end NICs (SPS-02)

1. In Windows Control Panel, click Network Connections.

2. Right-click Local Area Connection and then click Properties.

3. Under This connection uses the following items, click Internet Protocol (TCP/IP).

4. Click Properties.

5. Click Use the following IP address, and type in the IP address with your assigned front- end static IP address. (This is the IP address for the machine, not the cluster.)

6. Type in the subnet mask associated with this IP address.

7. Type in the default gateway with your assigned front-end gateway.

8. Click Advanced.

9. Under IP address, click Add.

10. In the IP Address box, type the cluster IP address as shown in Figure 12-5.

11. Type in the subnet mask associated with the cluster IP.

12. On the DNS tab, remove all DNS entries.

13. Deselect the Register this connections addresses in DNS box.

14. On the WINS tab, remove all WINS entries.

15. On the LAN Connection Properties page, under This connection uses the following items, click Network Load Balancing.

    

    Figure 12-5. Setting the cluster IP address (SPS-02)

16. Click Network Load Balancing again, and then click Properties.

17. On the Cluster Parameters tab, perform the following steps, as shown in Figure 12-6:

    1. In the IP Address box, type the cluster IP (number 1 on the worksheet).

    2. Type the subnet mask associated with the cluster IP.

    3. In the Full Internet Name box, type the DNS name.

    4. Change the Cluster operation to the Unicast setting.

         The Unicast setting can be used only when your server has multiple Network Interface Cards (NIC), while Multicast will be used when there is only one NIC. For a further explanation, please see the Windows Server 2003 Resource Kit or the “Network Load Balancing: Configuration Best Practices for Windows 2000 and Windows Server 2003” white paper.

    5. Leave Allow remote control check box unselected.

18. On the Host Parameters tab, perform the following steps, as illustrated in Figure 12-7:

    1. Because this is your second Web/search server, select Priority 2. (For additional servers, choose numbers in ascending order—3 and then 4, and so on.)

    2. In the Dedicated IP Configuration section, enter the address and subnet mask that you assigned to this machine in step 5.

       

       Figure 12-6. Configuring NLB cluster properties on SPS-02

       

       Figure 12-7. Configuring NLB host properties on SPS-02

19. On the Port Rules tab, in the Filtering mode section, verify that Multiple host is enabled and that Affinity is set to None as illustrated in Figure 12-8.

20. For large server farms: repeat these steps as necessary for each additional front-end Web server.

    

    Figure 12-8. Configuring port rules on SPS-02

Verify the Cluster’s Status

After configuring Network Load Balancing, it is absolutely paramount to test it and make sure it is working properly. Using Ping.exe and the Network Load Balancing Manager, you can quickly verify that your cluster is functioning properly.

1. In Administrative Tools, double-click Network Load Balancing Manager.

     If you run Network Load Balancing Manager from a node in the cluster, you might be presented with a warning that you need to use Multicast mode for successfully monitoring the cluster. If you did not select Multicast as the operating mode, you must run the Network Load Balancing Manager from a machine that is not in the cluster.

2. Right-click Network Load Balancing Cluster, and select Connect to Existing, as shown in Figure 12-9.

3. Enter the DNS of the cluster, and click Connect.

   Network Load Balancing Manager might take a minute or two to make all the connections. You should see all your servers in the cluster listed, and they should all be showing green, as shown in Figure 12-10.

   

   Figure 12-9. Connecting to a cluster in NLB manager

   

   Figure 12-10. NLB Manager listing of all servers in a cluster

4. From a computer not in the cluster, run ping –t <DNS name of cluster>, as shown in Figure 12-11.

   

   Figure 12-11. Starting the ping test for a cluster

To test to make sure the cluster is operational, issue a continuously pinging to the VIP of the cluster. The ping should not stop even though one of the cluster nodes is down.

To test the failover of the cluster, each machine in the cluster needs to be drainstopped, one at a time. Drainstopping is used to tell the server to continue processing existing requests but not accept any new requests, thus gracefully taking the server down.

Use the telnet command to test drainstop. Syntax: telnet <ip_addr_VIP> 80 or telnet <ip_addr_VIP> 443 , where <ip_addr_VIP> is the IP address configured in NLB for the cluster. As each node is drainstopped, the next node will accept the telnet session, but previously drainstopped nodes will ignore the telnet. This can be shown by using netstat -ano on all of the nodes in the cluster. The IP address of the client will not appear with a connection to port 80 or 443 on any drainstopped server, unless there are existing connections to that server.

To test failover, do the following:

1. Return to Network Load Balancing Manager.

2. Right-click the first server in the cluster (SPS-01), click Control Host, and then click Drainstop, as illustrated in Figure 12-12.

   

   Figure 12-12. Performing Drainstop on SPS-01

3. You will see the server icon turn yellow and its state set to Converging. Wait about 10 seconds, right-click the cluster name, and choose Refresh. The icon for the first node in the cluster (SPS-01) should be red now, which signals that it is stopped, as illustrated in Figure 12-13.

   

   Figure 12-13. SPS-01 converging after Drainstop

4. Look at the computer running the ping command and make sure it is still getting a response from the cluster, as shown in Figure 12-14.

   

   Figure 12-14. Ping test is still getting responses

5. Return to Network Load Balancing Manager.

6. Right-click the first server in the cluster (SPS-01), click Control Host, and then click Start, as shown in Figure 12-15.

7. You will see the server icon turn yellow and its state set to Converging. Wait about 10 seconds, right-click the cluster name, and choose Refresh. The icon for the first node in the cluster (SPS-01) should be green now, which signals that it is started.

   

   Figure 12-15. Restarting Network Load Balancing on SPS-01

8. Repeat steps 2 through 7 for each additional node in the cluster.

As long as the telnet and netstat indicate the client has made a successful connection to each node in the cluster, Network Load Balancing is configured properly. If one or more of the nodes fail to show a netstat session on port 80 or 443 from the test client, go back and reconfigure Network Load Balancing on each failed server as specified in the previous section.

Installing and Configuring SQL Server

SQL Server is required for a server farm installation or any installation that will have an extranet or Internet connectivity to the portal site. Running SQL Server on a separate machine from all other SharePoint Portal Server components is required, and it is recommended that you install SQL Server cluster if you want high availability. When installing on Windows Server 2003 Enterprise Edition, ensure that you are installing SQL Server 2000 and SQL Server Service Pack 3a. To verify that you have SQL Server 2000, on the CD, browse to \X86\binn\ssnetlib.dll. The version number should read 2000.80.311.0.

  For detailed instructions on how to install SQL Server 2000, please refer to the SQL Server 2000 Resource Kit.

You must properly configure SQL Server services before installing SharePoint Portal Server 2003. To configure SQL Server services:

1. Configure the MSSQLSERVER service and the SQLSERVERAGENT service to start automatically, and confirm that both services start successfully at startup.

2. Configure the service accounts for the MSSQLSERVER service and the SQLSERVERAGENT service to run as domain\spssql on the services corresponding to the instance of SQL Server you want to use.

3. For clusters, repeat this on each computer that is running SQL Server.

Installing Internet Information Services and SharePoint Portal Server

The SharePoint Portal Server client and administration functionalities depend on a specific Internet Information Services (IIS) configuration. Install IIS on all front-end search and index servers (SPS-01, SPS-02, SPS-03—and SPS-04 and SPS-05 for large server farms) as indicated in the following procedure.

To install Internet Information Services

1. Click Start, and then click Manage Your Server.

2. Click Add or remove a role.

3. Click Next to start the wizard. The Configure Your Server Wizard will gather information about your system, as shown in Figure 12-16.

   

   Figure 12-16. Starting the Configure Your Server Wizard

4. Click Application server (IIS, ASP.NET) from the list of choices, and click Next as illustrated in Figure 12-17.

   

   Figure 12-17. Selecting Application Server

5. Select the Enable ASP.NET check box.

6. Ensure that the FrontPage Server Extensions box is cleared. Click Next as shown in Figure 12-18.

   

   Figure 12-18. Setting custom configuration options for Application Server

7. Verify the install summary, and click Next to start configuration, as illustrated in Figure 12- 19.

   

   Figure 12-19. Verifying configuration options

8. Repeat these steps for additional servers (SPS-02, SPS-03).

Installing SharePoint Portal Server 2003

Install SharePoint Portal Server 2003 on all Web, search, and index servers (SPS-01, SPS- 02, SPS-03—and SPS-04 and SPS-05 for large portal farms). Please refer to Chapter 3 for detailed instructions on installing SharePoint Portal Server 2003.

Configuring the SharePoint Portal Server System Architecture

When you complete the SharePoint Portal Server 2003 installation, the SharePoint Portal Server Central Administration page displays. Table 12-4 shows which configuration tasks are prescribed for medium and large server farms. These tasks are described in subsequent sections. You should complete the appropriate configuration tasks based on the type of server and server farm you are configuring. When you finish configuring the server farm accounts and connecting to the configuration database, you should add the servers to the server farm topology.

Table 12-4. Tasks Prescribed for Medium and Large Server Farms

	Medium Server Farm			Large Server Farm
Configuration Task	First Front- end Web and Search Server	Second Front- end Web and Search Server	Index and Job Server	First Front-end Web Server	Second Front-end Web Server	Search Servers	Index and Job Server
Configure the server farm accounts	√	√	√	√	√	√	√
Specify the configuration database	√			√
Connect to existing configuration database		√	√		√	√	√

Configuring the Server Farm Accounts

Perform the steps listed in the following procedure and shown in Figure 12-20 on all the SharePoint Portal Server–based servers.

Figure 12-20. Configuring server farm accounts

To configure the server farm accounts

1. On the Configure Server Farm Account Settings page, in the Default Content Access Account section, click Specify Account.

     The Default Content Access Account can be any account, but to work properly, this account must have Read access to all content you plan on crawling.

2. In the Portal Site Application Pool Identity section, enter the domain and user name for the SharePoint Portal Server service account (domain\sqlsps).

3. Enter and confirm the account password.

4. Repeat these steps on each additional server.

Specifying the Configuration Database

SharePoint Portal Server uses a configuration database to store all information about the configuration and settings of a farm. This database needs to be created by the first server in your farm and used by each additional server you want to participate in the same farm.

To create the configuration database (SPS-01)

1. On the Specify Configuration Database Settings For SPS-01 page, perform the following steps, as illustrated in Figure 12-21:

   1. In the Database Connections section, click Create configuration database.

   2. In the Configuration Database Server section, in the Database server box, enter the name of the server (or virtual server if a cluster) running SQL Server 2000 (SQL-01).

   3. In the Configuration Database Name section, leave the default settings.

   4. Click OK. The Configure Server Farm Account Settings page appears.

   5. In the Contact E-mail Address section, enter the e-mail address of a system administrator.

   6. In the Proxy Server Settings section, set the appropriate proxy settings for accessing the Internet.

   7. Click OK.

      

      Figure 12-21. Creating the Configuration Database

        Typically, you should use the default database name. If you have a special circumstance—such as needing to name the configuration database according to a naming scheme your company set up—you can specify your own name for the configuration database. Just make sure to write it down, as it will be needed when installing each additional farm server.

To specify the configuration database (SPS-02, SPS-03 and SPS-04, SPS-05 on large farms)

1. On the Specify Configuration Database Settings For server_name page, do the following:

   1. In the Database Connections section, click Connect to existing configuration database.

   2. In the Configuration Database Server section, in the Database server box, enter the name of the server (or virtual server if cluster) running SQL Server 2000 (SQL-01).

   3. In the Configuration Database Name section, leave the default settings. (If a custom database name was specified, enter that here instead.)

   4. Click OK. The Configure Server Farm Account Settings page appears.

   5. In the Contact E-mail Address section, enter the e-mail address of a system administrator.

   6. In the Proxy Server Settings section, set the appropriate proxy settings for accessing the Internet.

   7. Click OK.

2. Repeat these steps for each additional server in the farm.

Adding Servers to the Topology

Now that all of the servers are added to the configuration database, you need to add them to the server topology.

To add servers to the server farm

1. On the Configure Server Topology page, click Change Components. Depending on the type of server farm you are adding, perform one of the following steps:

   For Medium Server Farms.

   On the Change Component Assignments page, in the Component Assignment section, there is a row of check boxes for each server and there are columns of check boxes to assign Web, search, and index roles. For each front-end Web server (SPS-01, SPS-02), select Web and Search. For the index server (SPS-03), select Index as illustrated in Figure 12-22.

   For Large Server Farms.

   On the Change Component Assignments page, in the Component Assignment section, there is a row of check boxes for each server and there are columns of check boxes to assign Web, search, and index roles. For each front-end Web server (SPS-01, SPS-02), select Web. For each search server (SPS-04, SPS-05), select Search. For the index server (SPS-03), select Index as illustrated in Figure 12-23.

2. In the Job Server Component section, select an index server from the Job server list.

3. In the Document Library Server Component (Optional) section, make sure the Document library server box is left empty.

   

   Figure 12-22. Assigning servers to the medium farm topology

   

   Figure 12-23. Assigning servers to the large farm topology

Specifying the Global E-mail Server

If you want portal sites in your farm to send e-mail alerts and e-mail invitations to other sites on those portal sites, you need to specify a global e-mail server for the farm, as outlined in the following procedure and illustrated in Figure 12-24.

Figure 12-24. Configuring Global E-mail Server settings

To specify a global e-mail server

1. On the Configure Server Topology page, click the name of the current global e-mail server (should currently be “unknown”).

2. Enter the Fully Qualified Domain Name (FQDN) of your e-mail server, an e-mail address to send from, and a reply-to e-mail address.

     The e-mail address you specify to have the e-mail sent from must have anonymous SMTP access. If your server requires authentication for the specified user, the portal site will be unable to send e-mail alerts.

Creating Portal Sites

Each portal site you want to have must be created separately on its own virtual server, and each will need to have its own URL. The following sections detail all the steps involved with creating new portal sites.

Creating Virtual Servers

Working in a load-balanced environment, the following IIS virtual server parameters must be configured the same on each front-end Web server:

• Virtual Web name

• Virtual Web IP address

• Application pool

• Virtual Web-file system directory name and location

To create the virtual servers for a new portal site

1. Open Administrative Tools and then click Internet Information Services.

2. Right-click the Web Sites container, point to New, and then click Web site.

3. Follow the directions to start the wizard. On the Web Site Description page, enter a description (such as “SharePoint Cluster,” which is used in Figure 12-25), and then click Next.

   

   Figure 12-25. Entering a description for the virtual server

4. On the IP Address And Port Settings page, select the IP address of the cluster in the Enter the IP address to use for this Web site box. Leave the default TCP port of 80 in the TCP port this Web site should use (Default 80) box, as illustrated in Figure 12-26.

5. On the Web Site Home Directory page, enter the path to the folders that contain the Web site for the virtual server, as illustrated in Figure 12-27. (This can be any physical file folder except Inetpub\wwwbin, because that folder is already used by the default virtual server.) Deselect the Allow anonymous access check box, and then click Next.

6. On the Web Site Access Permission page, verify that only Read and Run scripts (such as Active Server Pages [ASP]) are selected, and then click Next. Finish the wizard.

7. Repeat the preceding steps on each front-end Web server.

   

   Figure 12-26. Configuring IP and Port Settings

   

   Figure 12-27. Configuring the home directory

     For more information, see “Extending Virtual Servers” in the SharePoint Portal Server 2003 Administrator’s Guide, which is accessible from the Start menu.

Creating a Portal Site

Once IIS is installed and configured on each server running SharePoint Portal Server and the farm topology is set up properly, it is time to create portal sites on the farm. Each portal site has its own content and settings. Each portal site hosted on the farm can be considered to be a separate entity. If you use shared services, which are covered in Chapter 23, “Personalization Services in SharePoint Products and Technologies” you can share indexing, My Sites, and user profiles between portal sites to conserve resources. This section will describe how to create the first portal site in the farm, but the same procedure can be applied for each additional portal site to be created. (See Figure 12-28.) A portal site farm must contain at least one portal site.

Figure 12-28. Creating a portal site

To create a portal site

1. Click Start, point to All Programs, point to SharePoint Portal Server, and then click SharePoint Central Administration.

2. In the Portal Site and Virtual Server Configuration section, click Create a Portal Site.

3. On the Create a Portal Site page, verify that Create a portal is selected.

4. Fill in the Site Name of the Portal (in this example, Woodgrove Bank), select the virtual server created previously (SharePoint Cluster), and enter the account and e-mail address of the portal site owner. (See Figure 12-28.)

5. Click OK to continue. Click OK to confirm portal creation. A progress bar will display the status of the creation.

     If portal site creation fails, check the log file to determine the cause of failure. Correct the problem and try again. Often a creation failure is related to accounts not having proper permissions.

After you have created the portal site on the first load-balanced server, you must extend it to subsequent virtual servers.

To extend the portal site to the virtual servers

1. On the Operation Successful page, in the Server Extension Links section, click Link to a Virtual Server Extension Page for server-name (SPS-02), as illustrated in Figure 12- 29.

   

   Figure 12-29. Portal creation finished page

2. The Virtual Server List administration page for that front-end Web server opens. Click the name of the virtual server to which you want to extend the portal site (SharePoint Cluster) as illustrated in Figure 12-30.

3. This opens the Extend Virtual Server administration page. Click Extend and map to another virtual server.

4. On the Windows SharePoint Services Extend And Map To Another Virtual Server page, in the Server Mapping section, click the name of the virtual server the portal site was created on in the list.

5. In the Application Pool section, click Use an existing application pool, select the portal site application pool that was created previously in this chapter, and then set the application pool. Click OK.

6. SharePoint Portal Server will now contact each front-end server to update its configuration. You might be prompted for local administrator credentials for all the other SharePoint Portal Server–based servers joined to the topology. On the Refresh Config Cache On Other Web Servers page, click OK as illustrated in Figure 12-31. (You will see SPS-04, SPS-05 listed as well if you are performing a large farm installation.)

   

   Figure 12-30. Extending the portal site to other virtual servers

7. Repeat as necessary for each front-end server.

   

   Figure 12-31. Extending the virtual server completed successfully

     If the server has Internet Explorer Enhanced Security Configuration enabled, you must add each server’s administration website to Internet Explorer’s Trusted Sites list.

Now that the portal site is created, it should be verified to ensure it is functioning properly.

Open Internet Explorer. In the address bar, type the address of the portal site you created (for example, https://cluster.sps.test.local). It should look similar to Figure 12-32.

Figure 12-32. The successfully created portal site

Configuring Alternate URLs

The portal site has been created but is accessible only by the URL https://cluster.sps.test.local. This URL is the default portal URL and is used for crawling portal site content, but a different URL can be configured to allow easier access by end users. An alias record should be created in DNS for the URL end users will access, and that address should be specified in Alternate Access Mappings (shown in Figure 12-33) in SharePoint Central Administration.

Figure 12-33. Setting alternate access URLs

  If you do not know how to set up an alias record in DNS or do not have permissions to do so, you should contact your DNS administrator to get this set up.

  Extranets and Alternate Access Mappings will be covered in further detail in Chapter 13, “Installing and Configuring Windows SharePoint Services in an Extranet.”

To configure Alternate Access Mappings

1. Click Start, point to All Programs, point to SharePoint Portal Server, and then click SharePoint Central Administration.

2. In the Portal Sites and Virtual Server Configuration section, click Configure alternate portal site for intranet, extranet, and custom access.

3. Click the arrow next to the mapping name to configure, and then click Edit on the menu.

4. In the Intranet URL box, enter the address of the alias that was set up in DNS (https://woodgrovebank).

5. Click OK.

Summary

In this chapter, you learned about medium and large server farms. You should be able to plan a farm and successfully build that farm. The chapter also covered configuring and testing Network Load Balancing. You are now ready to implement additional portal sites and customize your main portal site.