Best Practices for Incremental Updates

Scope

This article describes best practices related to planning incremental updates, as well as the potential performance impact of large incremental updates.

Symptoms/Context

A detailed description of how to work with incremental updates is available in the Incremental Updates article. Performance impact caused by incremental updates can be avoided by following a few best practices.

Resolution

Best Practices

• Use the latest FAS version. The processing of incremental updates is constantly being improved. By using the latest version the latest fixes and improvements are available.
• The smaller in number of values and number of attributes an incremental update is, the faster it will be processed. The total time processing incremental updates however is lower when multiple updates are grouped in one input file. Try to get a good balance for this trade-off. Generally, for on-premise, try to get the number of updated nodes below 500 per update. This can be seen in the fredhopper.log. For Cloud customers, please contact Support.
• Use incremental updates only for the scenarios described in the Incremental Updates article. This is the intended usage.
• Timing is key as pushing a large incremental update (e.g. 500+ node updates) during your site's peak usage hours can cause hick-ups. FAS is modifying the index without a reindex, while also handling more queries than usual. We recommend pushing the incremental update after the peak usage time has passed, or pushing the updates along with a full data load and reindex at some other time.

Potential Issues

• Under circumstances, incremental updates trigger config reloads. This is the case when multi-lingual data used in campaigns has been changed. This can lead to increased response times for queries processed when the config is reloaded.
• Incremental updates are processed by the indexer and therefore this slows down also the Merchandising Studio.