This version of Infobright has zero support for distributed scalability. The internal smart grid employed for each table has a major flaw in that the data size cannot be expunged until 2GB of data is reached at the column-level.

This is a major flaw, making usage in a big-data scenario impossible. This means that you can delete as many records from a database table as you want. However, unless the 2GB aggregate size threshold was reached for some of the columns in the table, no reduction in disk space usage will occur.

Only the data from the columns that reached 2GB will actually decrease. Other columns below 2GB in size do not leave the disk.

I spent countless hours trying to find some workaround for this. I have nightmares of my e-mail inbox full of unsolvable questions about data size reduction from our field engineers.

It was rather slow, we have had made all reports at night to avoid overload. Any tables modification (except only adding new columns to tables) was prohibited. Even new columns must be added to the end of a table.

We ran into some quirks that Infobright had. We interacted with Infobright's support and were able to resolve them.

There still are issues with data replication - Infobright is currently for one server (unless you buy the Infobright appliance). This would mean that redundancy is something you need to implement yourself.

MPP, distributed processing!!! And better integration with Hadoop.

Additionally, further joins optimization.

Complex subquery and analitycal functions could be improved.

There is need for additional DDL functions.