Influxdb downsampling strategy

Question

We are using influx DB. Basically I wish to store my data for one year total so I created a default RETENTION POLICY fro 365d.

I also want to create downsampling. I store my event every second, but I dont need them in seconds resolution for more than 3 months. basically, after 3 monthes I prefer to SUM the data in days and after 6 months I prefer to sum the data in weeks resolution.

I understood from this doc that I can downsample using "continues queries".

What I also understood that I have to downsample to another table. This will make my query process much harder because I will have to decide which table should I query and maybe combine the data of several tables.

Can I somehow create Downsampling process that will have all the data in the same table?

Regards, Ido

AFAIK, this is not possible. Querying from different retentions will have to be expressed in the queries. — Wrikken
Thanks. Do you have an example of how to do this? Join? Union? — Ido Barash
No, what I am saying is that your 2 requirements (a) have different retentions / sample rates between now and >3 months ago and (b) keeping queries agnostic of that fact is impossible. What I would do is just decide which retention / precision I want to use: if I query over the last half year, just use the retention with the 'days' , for the last day the 'second' precision retention policy, etc. Keep in mind that with proper downsampling, your 'days' and 'weeks' for recent data are filled as well. — Wrikken

dannyman dannyman · Accepted Answer · 2018-06-04T22:08:13

To summarize the comments on this question: with Influxdb, you can not have multiple retention policies within a single table. You will have to query across multiple retention policies. You might want to Subscribe to this Github feature request for Intelligent Rollups and Querying of Aggregated Data

Influxdb downsampling strategy

1 Answers