Hi.

I have a question regarding query execution time.

I have one table with call log data. Total number of rows = 1,037,104,310

When I do aggregated functions like COUNT, MIN, MAX against entire data set, query execution is very fast = 0.0212s

However when I do COUNT GROUPED BY CALL CONNECT DATE, query takes about 200-300 seconds. Query returns only 27 rows.

Is that normal performance.

I was comparing query execution time against Vertica on single node database and when executing aggregate functions against entire database, Infobright is much faster. But when using WHERE or COUNT GROUP BY CALL_CONNECT_DATE, Vertica is much faster. As example COUNT GROUPED BY CALL CONNECT DATE or WHERE [CALL CONNECT DATE] =DATE in Vertica will take only about 15 seconds comparing to 200 - 300 second in Infobright.

I’m using community addition to evaluate Infobright performance.

The aggregates like COUNT,MIN,MAX are stored in what we call the Knowledge Grid, which is an in memory data node (so to speak) about the data in the database.  Each table and column is broken into what is called a data pack, these data packs have knowledge nodes associated with them that contain this information about the data.  So when the information you are looking for is in the knowledge grid, it is very fast.  When the data has to be resolved by opening up the data packs, like in the case of useing GROUP BY, then there is additional processing time.  However, in IEE, which is our enterprise version, these features are multi threaded and have a huge performance increase over ICE.

Is Infobright Enterprise Edition (IEE) Trial has this feature available?

Yes it does.  The Trial version of IEE is a fully functioning version.