SharePoint Portal Server 2003 depends on several key hardware resources to ensure optimal performance. In general, the most important resource for responding to increased load is CPU capacity.

 Note   Insufficient RAM, hard disk capacity, or network throughput can result in servers falling short of the ideal performance that their CPU capacity suggests.

Plan for hardware that delivers the CPU capacity and supporting resources that satisfy your requirements based on the information in this paper.

You can evaluate the throughput capacity of portal sites in many different ways. It is important to understand those throughput characteristics — Web page throughput, search throughput, and index throughput — that have the most significant impact on the performance of a portal site deployment. The following sections provide detailed recommendations for several different deployment scenarios.

Web Page Throughput

Web page throughput is a metric that can be hard to predict. Usage that drives Web page throughput can vary greatly from hour to hour and from day to day. SharePoint Portal Server 2003 is designed to provide a high-performance solution that can accommodate dramatically varying throughput needs. Conservative recommendations for capacity planning assume that, on average, the portal site deployment runs at 10 percent of total capacity. This enables the deployment to successfully respond to unusual high-demand periods.

There are many models and formulas for estimating the number of pages per second required to support a given number of users. However, it is not always clear what the term "number of users" means for an organization. It is common to refer to the number of users that could potentially use the portal site as the "number of named users."

It is also common to refer to the number of users that may actively use the portal site as the "number of simultaneous users." It is extremely difficult to make a reliable prediction of the number of simultaneous users, or the number of pages per second required to support them.

To help determine the required throughput of a portal site solution, organizations can use the following formula:

number of users	x	percent of active users per day	x	number of operations per active user per day	x	peak factor
360,000		number of hours per day

The following table explains the variables used in the formula.

Term	Definition
Number of users	The total number of users that may have access to the solution
Percent of active users per day	The percentage of the total number of users who might use the portal site solution during any particular day. Typically, this figure is approximately 25 percent, but it may vary from 10 to 75 percent.
Number of operations per active user per day	The number of operations that a typical user does on the portal site during a typical day. An operation is an action such as viewing the home page, searching, retrieving documents, etc. Typically, this number is approximately 10, but it may vary depending on the organization.
Peak factor	An approximate number that estimates the extent to which the portal site throughput exceeds the average throughput. This number typically ranges from 5 to 10.
Number of hours per day	The number of hours during which most activity occurs. This number typically ranges between 6 and 14 hours.

The number 360,000 is determined by:

100 (for percent conversion) x 60 (number of minutes in an hour) x 60 (number of seconds in a minute)

You can use these quantitative descriptions of a portal site deployment to estimate the required peak throughput. For example, a company with 10,000 users (of which 40 percent per day are active, performing an average of 20 operations) with a peak factor of 6 and 12 hours as the number of hours per day during which most activity occurs, needs 11 pages per second throughput.

10,000 x 40 x 20 x 6	= 11.11
360,000 x 12

Search Throughput

Use the total Web page throughput to estimate the number of content searches executed per second. Conservative recommendations for capacity planning assume that 10 percent of all Web pages viewed result in content searches.

Index Throughput

The rate at which content changes across your organization determines the rate at which to update the content index. In general, assume that 10 percent of the entire corpus must be updated in the index every 24 hours. While it is extremely rare for 10 percent of content to actually change every 24 hours, this recommendation allows the portal site to complete both large additions of content and strategic full updates of the index in a timely fashion.

The index throughput affects the performance of both the search and alerts mechanisms on the portal site, since these are dependent on an up-to-date index.

Recommendation

The index throughput should be capable of updating 10 percent of the entire corpus in the index every 24 hours.

High Availability

As a central source for important business information and applications, portal site deployments are frequently an important resource for an organization and can be classified as "mission critical." Many organizations choose to deploy a server farm to ensure portal site availability regardless of actual throughput requirements. Organizations typically deploy a server farm to ensure high availability rather than high Web page throughput.

Recommendation

If your organization classifies the portal site as a critical resource, consider deploying a server farm.

Storage

SharePoint Portal Server 2003 stores data in SQL Server and full-text indexes in the file systems on the search and index management servers. In general, the most important characteristics for determining the amount of storage space required are the total size of the documents stored on the portal site and the total size of the documents included in the portal site index. The following table illustrates the storage requirements for the server roles in a SharePoint Portal Server solution.

Server role	Required storage
Database	200% of the total size of all documents stored on the portal site
Index	60% of the total size of all documents stored on the portal site The index size is about 30 percent of the size of all documents in its catalogs. Because a copy (a snapshot of the content indexes) is always present, doubling the required space to 60 percent.
Search	Number of index servers X 60% of the the total size of all documents stored on the portal site When the index propogates from the index server to the search server, there are two copies of each index (each 30 percent of the total document size) on the search server for each index server.

For example, a portal site that stores 1 million documents with an average document size of 100 kilobytes (KB) stores 100 GB of document data and, thus, requires 200 GB of storage space.

Adding new portal sites or team sites does not in itself consume much disk space. Each new portal site (without content) consumes approximately 20 megabytes (MB) of disk space (in the database), whereas a new site, personal site, or portal site area (without content) consumes less than 200 KB of disk space (in the database).

Recommendation

Use the preceding table to compute your storage needs. Multiply the results by a factor of 1.5 to 3 to accommodate for future growth.

Other Recommendations

• For more than one portal site, use shared services. This results in a lower memory footprint, and requires no additional servers.
• Try to consolidate to a few portal sites, typically along geographic or organizational boundaries.
• Use areas for navigation and hierarchical security. In many cases, you can use areas to store divisional information rather than creating separate portal sites for divisions.