Multiple timestamp dimension

Apache Druid Q&A Google Groups
#1

Hello folks!

I saw this question already been asked in the past and I am wondering if such feature is now available in the newest version of Druid.

Problem

We are currently partitioning input data based on the **current_timestamp **column. Then we are using this column in the ingestion spec as

…

“timestampSpec”:{
“format”: “YYYY-MM-DD HH”,
“column”: “current_timestamp”
}

…

``

In the input data we have an other dimension called new_timestamp containing a different timestamp. Would like to query Druid and split the result based on the new_timestamp column.

I understand this that cannot be done while submitting the query, since segments are already partitioned by the current_timestamp. Using filters I can ONLY see new_timestamp data within current_timestamp segment, but it’s not what I want.

Is it possible to re-ingest current data specifying multiple columns in the timestampSpec?

Or better, Is Druid able to handle multiple time dimensions?

#2

Hi Flippo:

Are you talking about secondary timestamp ? https://druid.apache.org/docs/latest/ingestion/schema-design.html#secondary-timestamps

Thanks

#3

Actually I tried it already but I cannot split by this second timestamp. It is ingested as a Long and therefore Druid treats it as a normal dimension. I can only apply a filter on top of it but the primary timestamp is leading the query interval. Since my second timestamp could be very far in the past, I should query Druid always with an interval of 2 months and then apply the filter on the second timestamp column.

Either I got it wrong or this is not a “proper” secondary timestamp column.