Estimating the Required Number of Partitions for AI Search Service

Windchill AI Assistant Plugin > Advanced Actions > Estimating the Required Number of Partitions for AI Search Service

Depending on the volume of indexed data, you may need to adjust the storage capacity of the AI Search service to accommodate the data load.

The Windchill AI Assistant currently indexes only the textual content of documents. For more information, see Indexing Limits section. Each AI Search partition can index up to 45 GB of text data.

This topic explains how to estimate the number of partitions required based on the content type stored in your Windchill vault.

Estimation Method

When estimating the number of partitions, consider only documents that meet both of the following criteria:

• The document is of a supported file type (see Document Indexing Specifications).

• The document is accessible to the indexing user, as configured in the Indexing Dashboard.

As documents may contain images or other non‑textual content, only the textual portion of each file is indexed. Use the guidelines in the following table to estimate the amount of text content for each document category.

Document Type	Estimate of Actual Text in the Document
Plain text formats (such as TXT, CSV, log files, or source code)	100%
Text markup formats (such as XML, HTML, or Markdown)	100%
Word processing formats (such as DOCX, ODT, RTF)	10%–50% (lower for image‑heavy documents, higher for text‑heavy documents)
Spreadsheets (such as XLSX, ODS)	10%
Presentations (such as PPTX, ODP)	10%
PDF documents	10%

Calculating the Number of Partitions

To estimate the required number of AI Search partitions:

1. Determine the total size (in GB) of documents in each file type category.

2. Multiply the size of each category by its estimated text percentage.

3. Add the resulting text sizes to calculate the total estimated text data size.

4. Divide the total text data size by 45 GB.

5. Round the result up to the nearest whole number.

6. Optionally, increase the final number to account for expected future data growth.

Example

Assume the Windchill vault contains the following document categories:

Document Type	Total File Size (GB)	Estimated Text Percentage	Estimated Text Data (GB)
Plain text documents	17	100%	17
Text markup documents	24	100%	24
Word processing documents	13	25% (moderate image usage)	3.25
Spreadsheet documents	2	10%	0.2
Presentation documents	3	10%	0.3
PDF documents	15	10%	1.5

The total estimated text data size is 46.25 GB. Assuming a projected data growth of 20%, the estimated text data increases to 55.5 GB.

Dividing 55.5 by 45 results in approximately 1.23, which rounds up to 2.

In this example, the AI Search service requires two Standard S1 partitions.

For more information about tiers and pricing, see Azure AI Search pricing in Azure documentation.

Was this helpful?