Estimating SharePoint 2007 Index Server Requirements

The following information can be used to not only predict the initial size of the index, but can also be used in a modeling capacity to predict growth of the index based on growing content sources

(Pasted from http://technet2.microsoft.com/Office/en-us/library/9ccfb27f-ecba-4b7d-b9a0-88fac71478a31033.mspx?mfr=true)

Estimate disk space requirements

Use the following information to plan the disk space requirements for the index servers, query servers, and database servers in your environment.

Index server disk space requirements

To estimate the index server disk space requirements, we recommend that you use the following calculations:

• Size of data crawled = Y
• Size of index on index server = a range of 5% through 12% * Y = X
• Initial disk space = 2.5*X.

A large amount of index server disk capacity is required to accommodate backups, which must reside on the same disk as the index, and to accommodate the merge process when crawled data is merged with the index.

Note:

The volume of crawled data can differ based on the content source. A content source is a set of options that you can use to specify what type of content is crawled, what URLs to crawl, and how deep and when to crawl.

For example, if the content source specifies file-share content, the index size can be up to 30 percent of the size of the content.
You can estimate the size of the content index with the following equation:

Index size = Average size of document * number of documents * 4 x 10-10 GB.

Note that this equation is intended only to establish a starting-point estimate. Real-world results may vary widely based on the size of documents being indexed, and how much metadata is being indexed during a crawl operation.

Query server disk space requirements

Content indexes are propagated from the index server to every query server in the farm. The full index is propagated to the query servers during the query server initialization phase, and incremental changes in the index are propagated on a continual basis. The merging process requires more disk space than what is required to accommodate the index itself.

Given a content index size of X, we recommend that initial disk space be at least 2.5*X for every content index on each query server in the farm.