If only we always got things right the first time! In most cases, an eraser or bottle of white out will fail to fully address downstream realities we create by data management decisions we make today. The bottom line is that the indexes need to serve the application, and if they don't then something probably needs to change.

• "Oh, we should have tracked those details in the beginning..."

• "We've been told to change our product code format."

• "How come our logs are missing geographic information?"

• "What do you mean the data is not a number?!"

We re-index our Elastic search data for several reasons. When we're first trying ES, it's easy to whip up indexes that are not yet right. In this meetup, we can talk a bit about best practices and ways to prevent gotchas. BUT, the real aim of this meetup is not to lament past decisions, but rather to embrace the future. We want to reindex to help ES be suitable for new or evolving use cases. We like to make things faster when possible. AND, we love upgrading to promising new versions.

Two big mental thingamabobs in this discussion are: (1) the nature of working with immutable data itself -- even in such contexts as evolving EAV patterns and newly added sources of data, and (2) the recommended approaches that ES provides for reindexing (which is going to a new level in v5.0).