Hello,

I’m looking at possible data warehouse products for our company and am looking for some feedback from Infobright users/supporters. We are, generally speaking, collecting data from partners and clients into a WH, and using that to feed data to a web app for the same partners & clients. IE, this is not for our company’s own BI use. We currently have data in SQL Server (2005), with some ETL written in SSIS, and are just starting to understand how we want to present that data through a web app, ie what the business logic is, how much the data will need to be transformed, etc. So far it is looking like that data requests will be very predictable, so most of the crunching can be done beforehand, without needing to support intensive real-time drilling or data mining. We may stick with SQL Server as our only database, but I’m looking at Infobright IEE as something that might be more scalable, easier to maintain, and give better performance. We don’t have any direct prior experience with MySQL. Does it sound like IEE makes sense for us? Has anyone ported a WH and/or ETL from SQL Server (SSIS) to Infobright?

Thanks for any feedback.

Harvey

I found tables optimized for Infobright perform very fast.  I try to make my tables so no joins are necessary in queries and they are pretty fast.  You should simply load the same tables in MSSQL and infobright and do queries.

Not having to maintain indexes is a definite plus.  But you will have to prejoin data (make the table conform to the query).  Prejoining for some may not work well, but for us I think we will be using it.

Ankur

Hi !

Surely you should try IEE. I understand there is an initial difficulty of getting familiar with MySQL family of tools and sometimes with restricted supported SQL syntax, but the possible benefits you have mentioned by yourself
The fact is that IEE was aimed rather at ad-hoc querying. A heavily tuned SQL server with proper indexes and aggregates can perform faster for a set of predefined queries. But with IEE you have the advantage, that if the user introduces new queries, you do not need to re-tune the database.

But even with IEE you can optimize the system towards expected set of queries by presorting the data to be loaded. IEE benefits the loaded data are presorted according to the most often used columns in query conditions. In does not mean you need to sort whole tables. If you load e.g. 1 mln rows at a time just sort each such chunk.

Cheers