Does InfluxDB IOx builds any indexes itself on its tag-columns excepts for native index in parquet?

InfluxDB IOx
1

I am trying to dive into the sources of InfluxDB IOx.

As `tags in InfluxDB are frequently considered as filtering conditions, however, they are stored as string-type columns as any other field columns (which is meant to store sensor data).

I used InfluxDB 2.x in my project, so I am really curious about whether InflxuDB IOx builds any indexes on those tag columns. Except for those built by Parquet itself (a.k.a. PageIndex).

I would appreciate any advice.
(This question had been posted on StackOverflow: time series - Does InfluxDB IOx builds any indexes itself on its tag-columns excepts for native index in parquet? - Stack Overflow)

2

Currently, it does not. With they way that the IOx query engine storage tier work and are structured, the indexing strategies from InfluxDB 1.x and 2.x don’t offer the same benefit in 3.x (based on IOx). We are in the process of implementing other performance enhancing features however, such as:

  • Custom storage partitioning
  • Query caching
  • Tiered storage

…and more.

1 Like