My requirement: I want to create a single table where I can do aggregations like sum on integer values of a column; but also hold string values in the same column. What I am dealing with: I have a realtime table where I am ingesting data with fields

timestamp

,

id

and `value`; the

id

describes the

value

being ingested; here

value

can hold both

INT

and

STRING

for my use case. I wanted to know which of these options are feasible: 1. Create columns

value_int

and

value_string

and use a filtering function in Pinot that can save records in

value_int

if

value

is

INT

, and vice-versa for

STRING

values of

value

. I tried this, but the filter function as shown in the docs, does not allow this 2. Store all values as

STRING

and use a pinot-specific

CAST

or

CONVERT

function to do aggregations. But I could not find a cast/convert function in Pinot. Thus I am not able to do

sum

operations on the data I would welcome any ideas/workaround for the same.

Either option works, it depends on your latency/sla/cost requirements on which one would be better. While 1 is better read optimized than 2, 2 avoids additional storage at the cost of read time computation.

Thank you.