Schema Chnages : Re-Indexing

Hi Team,

We were having schema like below (around 15/16 dimension and 1 metric, name of the metric is going in the dimension).

Metric name and value are going as name value pair against all the dimensions.

DeviceID (dimension)

Location (dimension)

IP_Address(dimension)

metricname(dimension — cpu, memory,availability,reachability, etc)

metricvalue (metric)

Is there any way we can change the schema with re-indexing like below [ basically I want to keep all the metrics as columns instead of name value pair ]

DeviceID (dimension)

Location (dimension)

IP_Address (dimension)

Hi David - Would you be able to check this once . Thanks in advance.

Hey Arpan,

As far as I know, if you can re-read the raw data, you should be able to make creative use of the Javascript parser to generate the fields that you want, see: http://druid.io/docs/0.9.2/ingestion/data-formats.html. Keep in mind though that the Javascript parser is quite slow. For best performance, you should have your data transformed by another step in your ETL pipeline before feeding it into Druid.

Also as far as I know, there’s no way to transform the data in this way once it’s been ingested. If you don’t have access to the raw data, you may need to extract the data and re-ingest it.