Plan for redundancy (Office SharePoint Server)

Applies To: Office SharePoint Server 2007

This Office product will reach end of support on October 10, 2017. To stay supported, you will need to upgrade. For more information, see , Resources to help you upgrade your Office 2007 servers and clients.

Topic Last Modified: 2016-11-14

In this article:

• About redundancy

• Define server redundancy requirements

• Plan for a limited server deployment

• Plan for a minimum level of server redundancy

• Choosing a baseline server farm topology

• Plan Web server redundancy

• Plan application server redundancy

• Plan database server redundancy

• Evaluating the risks of application server failures

• Select a baseline topology

This article describes the options for scaling out redundant server roles included in a Microsoft Office SharePoint Server 2007 farm. After reading this article, you will be able to identify and record the redundancy options that are appropriate for the environment.

For more information about availability, see Plan for availability (Office SharePoint Server).

About redundancy

The term redundancy is often misinterpreted to be synonymous with availability. While these concepts are related, they are not the same. Redundancy refers to the use of multiple servers in a load-balanced environment for any of several purposes, such as to improve farm performance, to scale out to accommodate additional users, and to improve availability.

Availability is a more specialized concept that refers to a multiple-server environment that is designed to accept connections and operate normally even when one or more of the servers in the farm are not operational. Therefore, availability implies redundancy, and additionally implies a failover mechanism and several other possible characteristics. A redundant system, however, might not be highly available.

This article describes how to implement redundant servers in an Office SharePoint Server 2007 farm.

Define server redundancy requirements

Office SharePoint Server 2007 supports scalable server farms for capacity, performance and availability. Typically, capacity is the first consideration in determining the number of server computers to start with. After factoring in performance, availability also plays a role in determining both the number of servers and the size or capacity of the server computers in a server farm.

By the end of this section, you will be able to decide if you need to build expandable capacity into the server deployment topology by deploying redundant servers (three or more servers), or if it makes sense for the organization to plan for a limited server deployment that has no redundant servers.

Plan for a limited server deployment

If you do not need to build additional capacity and performance into the server deployment, the starting point for the server topology is one or two servers. For a limited-use purpose, you can deploy a single server.

Limited-use purposes include the following:

• Installing Office SharePoint Server 2007 for evaluation purposes.

• Deploying only Windows SharePoint Services 3.0.

• Deploying a subset of the Office SharePoint Server 2007 features.

• Deploying Office SharePoint Server 2007 for a limited purpose (such as for a single department) or for a limited number of users.

The recommended starting point for most Office SharePoint Server 2007 deployments is at least two server computers:

• Server 1: Front-end Web server and application server computer

• Server 2: Dedicated SQL Server computer

  

If you have determined that you do not require server redundancy in the environment, you can now go to the following article to complete the next planning step: Plan for performance and capacity (Office SharePoint Server). The completion of this planning step will determine the total number of servers recommended for the server deployment plan. You do not need to read the rest of this article.

Plan for a minimum level of server redundancy

To deploy a redundant solution, you must deploy a server farm.

There are several different server topologies that can be used as a baseline. Each of these topologies builds in a level of server redundancy. This section provides an overview of these server farms.

Three-server farm

There is another alternative for achieving redundancy while deploying fewer servers. With a three server farm, you must choose which of the server roles to make redundant: either the Web server role or the database server role.

By adding a third server to the Web server tier, you achieve redundancy of the Web server role. The query and index roles can either be installed on the same Web server (see Option A below), or they can be installed on different Web servers (see Option B below).

Given this topology, the query role cannot be deployed to both Web servers to achieve redundancy. This is because, if the query server role is installed on the same server as the index server, the index server will not propagate the index to other query servers. However, you can install the index role on the database server, allowing you to deploy the query role on both Web servers. However, the performance of the database server will be affected.

Although availability is limited, dedicating two servers to the Web server role increases the overall performance of the small farm. Use this topology when performance is more important than data redundancy.

Four-server farm

The smallest server farm that builds in redundancy consists of four servers:

• Servers 1 and 2: Web servers and query role installed on both computers. Additional application server roles, such as Excel Calculation Services, can be installed on one or both servers.

• Servers 3 and 4: clustered or mirrored database server.

With a four-server farm, you must carefully choose where to deploy the index server role. The query role cannot be deployed to both the index server and another server in the farm to achieve redundancy. This is because, when the index role is installed on the same server computer as the query role, the index role no longer propagates content indexes to other query servers. Consequently, if you install the index server role to one of the Web servers, you lose the ability to host the query role on both Web servers. You can install the index role on the database server, achieving redundancy of the query role on the Web servers. However, the performance of the database server will be affected, particularly when content is being crawled.

Additional application server roles, such as Excel Calculation Services, can be installed on one or both Web servers, depending on performance and availability requirements. In the previous diagram, additional application roles are deployed to only one of the servers in both Option A and Option B. In Option A, performance is optimized if the Web server that is not hosting the combined index and query roles hosts the other application roles. In Option B, the topology is optimized for user requests and queries. Performance of the other Web server can be managed by scheduling indexing jobs during off-peak hours, enabling greater performance of the other application server roles during peak business hours.

Five-server farm

The most common redundant server farm topology introduces a middle tier and consists of five server computers.

Given this topology, you can install all application server roles on the dedicated application server. This design optimizes the performance of the front-end Web server computers by enabling you to offload one or more application server roles to the middle tier.

The primary planning decision to make with this topology is where to install application server roles. The index server role should remain on the dedicated application server. However, the decision about where to install additional application server roles depends on whether you want to optimize the server farm for capacity or for performance. If redundancy of application server roles is a priority, you can install application server roles that are designed to be redundant (Excel Calculation Services, Search query, and Microsoft Office Project Server 2007) on the two Web servers. To optimize for performance, consider moving first the Excel Services role to the application server and next the query role.

Deploying the query role to the Web servers is a common configuration that achieves redundancy of this role and can also optimize the performance of the overall server farm. The planning choices that you make after reading “Plan application server availability” later in this article will determine which option is most appropriate for the environment.

Six-server farm

For maximum redundancy with a minimum number of servers, deploy an additional application server to the middle tier for load balancing application server roles that are designed to be redundant. This server farm topology consists of six servers. The query role is installed to the front-end Web servers to achieve redundancy.

If you plan on deploying the Excel Calculation Services application server role or the Office Project Server 2007 application role to the farm and redundancy of these services is a priority, this is the recommended starting point. This topology protects these server roles from direct user connections and optimizes the performance of the overall farm when compared to smaller farms.

Choosing a baseline server farm topology

Each of the server farm topologies described earlier in this article represents a baseline starting point for designing the deployment. The starting point that best suits the organization depends on the server roles for which you require redundancy.

The rest of this article describes the redundancy options for each of the server roles. By the time you are finished with this article, you will be able to identify the baseline topology that can deliver the redundancy that the organization requires. This is the topology that you will use as a baseline when you start planning for capacity and performance.

Plan front-end Web server redundancy

Use this section to:

• Determine if the organization requires redundancy built into the Web tier.

• Plan which Web server load balancing technology to implement.

Most organizations require redundancy at the Web tier. There are a small number of scenarios in which a three-server farm with one server running the Web server role makes sense.

The next step is to plan which load balancing technology to implement. Office SharePoint Server 2007 supports two methods of load balancing:

• Software, such as Network Load Balancing (NLB) services in the Microsoft Windows Server 2003 operating system. NLB runs on the front-end Web servers, and uses TCP/IP to route requests. Because NLB (and other software load balancing solutions) runs on the front-end Web servers, it uses the front-end Web system resources, and thereby reduces the resources you can use for serving Web pages. However, the impact on system resources is not great, and a software solution can handle up to 32 front-end Web servers. For more information about NLB services in Windows Server 2003, see Network Load Balancing Clusters (https://technet.microsoft.com/en-us/library/cc759510.aspx). For more information about NLB services in Windows Server 2008, see Network Load Balancing (https://technet.microsoft.com/en-us/library/cc732855.aspx).

• Hardware, such as a router or a switch box. Load balancing hardware uses the network to direct Web site traffic between the front-end Web servers. Load balancing hardware is more expensive to set up than software, but it does not affect resources on the front-end Web server resources. Office SharePoint Server 2007 can be used with any load balancing hardware.

  We recommend that you set the load balancing affinity to None to improve availability. If you have a custom topology requirement, you may want to configure the affinity differently.

Although not recommended, there is a third method of load balancing, round-robin load balancing with Domain Name System (DNS). Round-robin DNS load balancing can use significant resources on the front-end Web servers, is slower than either load balancing software or hardware, and is not recommended for use with Office SharePoint Server 2007. Also, round-robin DNS load balancing does not take session load into account when routing a user to a server, which can lead to a server being overloaded.

Plan application server redundancy

Which baseline server topology you choose depends on the requirements for redundancy of application server roles. This section describes the application server roles relative to their redundancy options.

Application server roles for Office SharePoint Server 2007 fall into two categories:

• Roles that can be redundant

• Roles that cannot be redundant

Roles that can be redundant

These application server roles can be deployed to multiple servers. The code that is deployed to each server is identical and the application server roles do not store any data. In other words, each instance of these server roles remains identical. If one of the server computers fails, no saved data is lost. The Web servers automatically load balance requests to these server roles across the available application server computers.

The following application server roles can be deployed redundantly:

• Query.   The query role can be deployed to any number of application server computers or it can be deployed across Web servers. There is one limitation, however. If the query role is deployed to the same server that hosts the index role, the query role should not be deployed to any other server computers. This is because the index role recognizes that the query role is on the same server and, consequently, does not attempt to propagate the index. In some scenarios, you can optimize the throughput of the server farm by deploying the query role across the Web servers, even if you have introduced a middle tier for dedicated application servers. For example, if more than 50% of the content requests coming into the server farm are requests for static content, performance is potentially optimized by hosting the query role on the Web servers. This is because the query role caches the content that it serves, making it readily available for subsequent requests. The query role is unique in this sense.

• Excel Calculation Services.   The Excel Calculation Services role performs Excel calculations on Excel workbooks that are stored in the content databases. This application server role is unique in that it stores session-state information throughout the duration of a user session. When a user opens a workbook, the Web server role continues to route the user requests to the same Excel Calculation Services application server until the workbook is closed and the user finishes the session. The Excel Calculation Services role can be a resource-intensive role. In some scenarios, you can optimize the performance of the farm by deploying this role across all Web servers.

• Office Project Server 2007.   Although Office Project Server 2007 is not included with Office SharePoint Server 2007, it is built on the Windows SharePoint Services 3.0 platform and can be deployed to a Windows SharePoint Services 3.0 or Office SharePoint Server 2007 farm. The Office Project Server 2007 application role serves requests for Office Project Server 2007 services and data to the Microsoft Office Project Web Access servers.

Roles that cannot be redundant

Application server roles that cannot be redundant include index and Windows SharePoint Services 3.0 search. These application server roles can be deployed to multiple servers; however, the multiple servers are not redundant. These server roles are configured to crawl content and generate content indexes. If you deploy these roles to multiple servers, each server crawls different content.

• Windows SharePoint Services 3.0.   The Windows SharePoint Services 3.0 search application role is an option if you are not using Office SharePoint Server 2007 query and indexing. The Windows SharePoint Services 3.0 search application role includes both the search and indexing components. These components cannot be divided. Additionally, Windows SharePoint Services 3.0 search is required to provide full text search of Help. Consequently, you can consider deploying this application server role to a Office SharePoint Server 2007 farm to achieve this functionality (even if Office SharePoint Server 2007 search is deployed).

• Office SharePoint Server 2007.   In Office SharePoint Server 2007, the index role is associated with a Shared Services Provider (SSP). The index role builds one index per SSP. One index server can be associated with multiple SSPs. However, indexes across SSPs cannot be combined. You can deploy multiple index servers to improve capacity. In this case, each index server is associated with different SSPs. Unlike the Windows SharePoint Services 3.0 search role, content indexes produced by the Office SharePoint Server 2007 index role are continuously propagated to all servers that host the query role in a farm. Consequently, the output of the Office SharePoint Server 2007 index server role (that is, the index) is considered redundant if the query role is deployed to more than one server computer.

If you are deploying a Office SharePoint Server 2007 farm, we recommend that you use the Office SharePoint Server 2007 query server and index server roles. This enables you to scale out the query component, achieving redundancy of the content indexes. Because the primary reason for deploying multiple Office SharePoint Server 2007 index servers or Windows SharePoint Services 3.0 search servers is to scale for capacity or performance, subsequent planning articles will help you decide if multiple servers are recommended for your deployment.

The following table indicates which application server roles can be deployed redundantly and which roles can be deployed to multiple servers but are not redundant.

Plan database server redundancy

Use this section to help you determine if redundancy of the database server role is a requirement for the solution. Subsequent planning topics will help you decide which database redundancy technology is most appropriate for the environment. For more information, see Plan for and design database storage and management.

The database server role affects the availability of your solution more than any other role. If a Web server or an application server fails, these roles can quickly be restored or redeployed. However, if a database server fails, your solution depends on restoring the database server. This can potentially include rebuilding the database server and then restoring data from the backup media. In this case, you can potentially lose any new or changed data dating back to the last backup job, depending on how SQL Server 2005 is configured. Additionally, the solution will be completely unavailable for the time it takes to restore the database server role.

Evaluating the risks of application server failures

This section summarizes the expected consequences of a single application server failure. In other words, if you deploy an application server role to just one server and the server fails, what are the potential consequences? Understanding the potential consequences will help you prioritize the allocation of servers in the farm. The following table lists application server roles and describes the consequences of downtime for each.

The general redundancy recommendation is to plan to install an application server role to at least two application server computers if:

• The solution is primarily based on the features provided by the application server.

• The availability requirement for the features provided by the server role is 99 percent or greater.

If your organization can tolerate temporary loss of this functionality for the amount of time it takes the IT team to deploy an application server role to a different server or to restore service to the existing server, consider deploying the role to a single application server.

Select a baseline topology

After you identify the redundancy requirements for the individual server roles, review the baseline server topologies and choose the topology that is most appropriate for the environment.

Download this book

This topic is included in the following downloadable book for easier reading and printing:

• Planning and architecture for Office SharePoint Server 2007, part 2

• Planning an Extranet Environment for Office SharePoint Server

See the full list of available books at Downloadable content for Office SharePoint Server 2007