Best Practices for Improving Search Performance

Search performance depends on many factors, including the size of the search index, your search settings, the search syntax you're using, the privileges you have as a user, and the hardware hosting the Laserfiche Full-Text Indexing and Search Service. If you're encountering slow searches, read on to learn how to change these factors to make your searches faster.

Search Index: What to Include?

The larger a search index is, the slower searches will be. Here are a few ways to reduce the size of your search index.

Data Types

Certain types of data are always indexed, while other types are not indexed by default. To improve search performance, check if there are data types being indexed that you can afford to make unindexed.

Example: One of your metadata fields is a text field that indicates the status of a document in a review process. It may contain short, easy-to-search-for phrases like "New", "Reviewed", or "Review in progress". To reduce the size of your search index, you may want to make this field unindexed. In contrast, you may have another field called "Comments", where users may leave several sentences of comments on the contents of the document. A field like this is a better candidate for indexing, as it may benefit more from features of indexed searches like fuzzy search and root word search.

New Documents

You can choose whether to always index new documents. If you plan to upload many documents that do not need to be indexed, you can turn off automatic indexing of new documents in the Laserfiche Administration Console. If a user later wants a specific document to be indexed, they can manually index it from the web client or Windows client.

Volumes

If you have old volumes that you're no longer adding documents to, and you do not need these volumes to be indexed, you can create a new search catalog that only indexes newer volumes. This reduces the size of your search index and can increase search speed. Your newer volumes can continue to accept new documents, and you can continue to auto-index these new documents.

To un-index older volumes

1. Detach the current search catalog.
2. Create a new search catalog.
3. Re-index the volumes that you want indexed. Do not index the repository.

Note: Documents that are added to the unindexed volumes will not be automatically indexed. Any documents that you want to be indexed in those volumes will need to be manually indexed.

Search Settings

In the web and Windows clients, you can configure search settings that may slow down or speed up your searches.

You can increase search speed by disabling the following settings:

• Find partial matches when performing a quick search
• Include variations of root word in indexed searches
• Use fuzzy search
• Resolve shortcuts

In addition, you can limit the number of results returned for indexed searches .

• In the client's search settings, select Limited indexed searches to and specify a number.
  • Applying any kind of limit can speed up searches on indexed text.
  • The lower the limit, the faster your indexed searches. However, lowering the limit may also eliminate search results that you want.

Optimizing Search Syntax

Due to the way the search engine parses advanced search syntax, some types of advanced searches are slower than others.

• For wildcard searches, search terms that end with a wildcard (e.g. term*) are faster than search terms that start with a wildcard (e.g. *term). This applies to all wildcard characters, not just *.
• For in-folder searches, searches can be slow if the folders in question contain a large number of entries. You can insert a special option into the Laserfiche Server database to speed up in-folder searches. Be aware that this can slow down in-folder searches if the folders have very few entries.

Columns Displayed

You can speed up searches by reducing the number of columns displayed in search results. In addition, certain columns contain information that is expensive to calculate. These columns are good candidates for hiding.

• The Location column is expensive to calculate because the file path has to be calculated for each result.
• The Security column is also expensive to calculate.

User Privileges

There are two privileges that can affect search performance:

• Bypass Browse. This privilege lets a user browse the repository even if they don't have Browse entry access right for the relevant entries. This speeds up searches because Laserfiche does not have to calculate entry access rights before displaying search results.

Warning: If you assign a user the Bypass Browse privilege, they'll be able to know of the existence of any entry in the repository just by entering an appropriate search term. If this poses a security risk, you should not assign them this privilege.

• Bypass Folder Filter Expressions. Checking for folder filter expressions slows searches down. If you're not using folder filter expressions, you can assign the Everyone group the Bypass Folder Filter Expressions privilege to speed up everyone's searches.

Hardware

The hardware you use to host the Laserfiche Full-Text Indexing and Search Service influences search performance. For best performance, try the following tips:

• Host the Full-Text Indexing and Search Service on a separate computer from the Laserfiche Server.
• If you have multiple repositories on a Laserfiche Server, try pairing each repository with a dedicated installation of the Full-Text Indexing and Search Service on a separate machine. For example, if you have three repositories, you would have three separate machines, each hosting a Full-Text Indexing and Search Service for one repository, and a fourth machine hosting the Laserfiche Server.
• Upgrade the CPU speed for the Laserfiche Server's SQL Server.