Perofmance impact of updating points frequently

Hi,
I have a scenario where I have multiple fields(say 5) in a measurement. But when writing points into this measurements I may not have all the fields and some fields may come later(say 2) with same tags and timestamp. The fields are related to each other but sometimes they are not received or revceived later. Now, I know InfluxDB will update the existing points if we write missing fields later with same tags and timestamp. Is uptating points frequently going to hit write performance significantly?

Hello @aateek,
If you “update” a series with the same series key at the same time, then you’re not really performing an update. The old series is dropped and replaced with the new one. InfluxDB is optimized for high writes and reads so this shouldn’t be a problem. These design considerations might be interesting to you.
https://docs.influxdata.com/influxdb/v1.8/concepts/insights_tradeoffs/

@Anaisdg Thanks for taking out time to answer my question. I had seen the link provided.
Actually, I have new points and updates with new fields for some of these new points in the same batch (I am batching points). So when writing through HTTP, both new points and updates are going in a single batch. I was guessing it should not be a problem but needed advice. I assume by “at the same time” you mean- within a very short period, say 1s.

1 Like

@aateek,
No I mean if you’re writing data at the exact same timestamp.

While it’s technically true that you can overwrite the existing measurement, it’s important to understand the particular use case when someone asks whether they can update data in InfluxDB.

When many users ask that question, they are thinking about it in the context of a relational database and CRUD operations, where updates are frequent, and for which InfluxDB is not well suited.

In general, time series data is composed of measurements taken at various point in time. It would be unusual to change the value of a measurement after it was taken. If you record the temperature in a room at 10:00am, for example, that value would remain static.

If you want to change the tags for a measurement, it usually makes sense to create a new series with the updated information and then drop the original series.

This topic was automatically closed 60 minutes after the last reply. New replies are no longer allowed.