About the Index Self-Optimization Process

The index optimization is a process during which Coveo Enterprise Search (CES) consolidates the index content. The optimization process recycles orphan identifiers from deleted documents and terms to optimize the index structure for faster queries.

The Coveo Platform 7 index features a self-optimizing process that continuously and automatically manages this index compaction. The self-optimization process does not require additional index hard disk space.

The self-optimization index process performs the following tasks:

• Monitors identifiers (DocID, PersitentDocID, TermID) that become available when documents or terms are deleted from the index.

• At each transaction, when a minimum number of orphan identifiers to recycle are available:

  • For each identifier, clears any remaining references to the identifier and reassigns the identifier to the pool of identifiers that can be used for new documents/terms.

    The identifier recycling respects processing time and percentage limits to prevent overloading the indexing process. The limits are less restrictive during off-peak hours, allowing for a more aggressive self-optimization process when more Coveo server resources are available.

    Note: You can define the off-peak hours for your Coveo system (see Modifying System Schedules).

  • Issues a message in the CES logs indicating the percentage of recycling performed.

Note: With the assistance of the Coveo Support, it is also possible to fine tune the behavior of the self-optimizing process by modifying the default value of associated parameters in the CES configuration file ([Index_Path]\Config\Config.txt).

In the case of Coveo instances with distributed indexes, the self-optimization process runs independently on each slice and on each mirror.

You can monitor the self-optimizing index process by looking at its messages appearing in the CES Console (see Using the CES Console