Would you expect lengthening a retention policy to have any effect on the typical performance of writes, or of queries?
We recently started a database with a small retention policy (one week) and were happy with that, but ultimately wanted more history, so we increased it to 6 weeks. Around the same time, we started noticing regular slowdowns (on a 10-minute period) in write latency, and the problem has gradually worsened.
My mental model of InfluxDB is that having more shards in the database shouldn’t really slow down writes in a big way, not should it slow down queries as long as they have a reasonable time bound on them. In fact, our database gets few queries but a lot of writes, and the writes are the main operation where we’re concerned about slowdown.
The association between expanding the retention policy and the write slowdown may be a coincidence, so I’m just looking for information whether it’s a worthwhile hypothesis to pursue.