Chunk_target_size vs. max_chunk_age

Hello,

I’m trying to tune Loki performance and I’m following the best practices documentation.

According to that I should aim to fill up chunks to chunk_target_size. As far as I know, the two clear options I have to achieve that is to

  1. adjust “sharding” i.e. add or remove labels to make streams of my desired log volume
  2. I can just increase max_chunk_age so chunks will not be flushed prematurely

What are the possible downsides of just increasing max_chunk_age? Increased memory usage? It would also take longer for data to be flushed to in my case S3. Which means queries would have to look for longer periods of data in the ingesters?

Any insight would be greatly appreciated :slight_smile: