How should i approach ID/string columns that are g...
# troubleshooting
g
How should i approach ID/string columns that are going to be only returned from queries and not used in any filtering/aggregations, for example i have TeamId and TeamName i have invertedIndex on TeamId, but TeamName will be used only as a result to some queries and i don't really need index on it, in which tableIndexConfig -> columns array i should put it? noDictionaryColumns or varLengthDictionaryColumns?
x
You can try the query with both teamId and teamName and check the perf with only teamID. If eventually the perf diff is too much, you can create a dimension table with just teamID to teamName mapping and use lookup join in the end.
g
i would have the same issue i want to minimize the penalty of storing the strings as well as the query will be something like select otherIds/names, teamId, teamName where teamId = 123 the queries won't be statistical aggregations, they will be mostly like a history
i wanted to go the route of dimension tables but didn't find a way to insert 1 kafka message inside a multiple tables, such that i can use the stream to populate multiple tables and have them synced
x