One of the things I am finding working more and more with Impala… specifically creating views and tables with impala sql… is that I have to be very specific when I am developing.
Impala sql is case sensitive. This means that if you capitalize “RowCreateDateTime” initially, every time you refer to that field it must be with the exact same case (upper and lower texts). Thus, “rowCreateDateTime” doesn’t refer to the same field as above because the r is lowercase in this instance.
Additionally, while the datatypes are supposed to be cast automatically, I find that it is not consistent. I am now specifically casting any non-string field specifically. I am also trimming each field before casting as I find the ingestion techniques that the ingestion team is using tends to pad some records/fields with lots of spaces at the beginning or end. Trimming takes care of that.
I am becoming more and more comfortable with using Impala… especially through the Hue web interface.
What do you think of Impala?
John