Tony Bain

Building products & teams that leverage data, analytics, AI & automation to do amazing things

Positioning your Database Start Up for Data Warehousing

July 06, 2009 | Tony Bain

BI/Data Warehousing is an easier market to enter for new database platform vendors.  This is for a few reasons.  Firstly, most BI deployments are custom built projects for each organization.  This means the ability to pick and choose various layers of the stack is much greater. 

Secondly, BI/DW projects success/failure metrics are often tied to database platform driven properties – performance, scalability, load times etc.  The ability to stray outside any existing database platform “standards” to choose a platform that better meets key metrics is more tolerable.

Thirdly, because the ratio of BI to OLTP is low, the associated impact of violating a corporate standard is much lower.  With OLTP applications typically deployed in the hundreds or the thousands within the enterprise, lack of firm standards could end up with dozens of different database platforms requiring operational support, spread across hundreds of systems.  On the other hand, violating the standard for a handful of DW systems is unlikely to turn into the management nightmare that would occur with the former situation.

The data warehouse database platform has been an area of heavy innovation and many newcomers have appeared over the last 5 years.   If you are going to enter this space you better make sure you have your point of difference pitch really honed. 

In addition you should:

  • Ensure you are supporting standard interfaces (OLE-DB or ODBC).  Being the greatest data warehouse platform that your customers can’t write reports for isn’t going to be a great sell.
  • Ensure you are providing a good standard set of tools.  Query tools, design tools, data loading/integration tools etc.  If you are small getting any of the third party tool vendors to pay attention to you is going to be difficult.  Use compatible interfaces where you can to ease this but also make sure include your own support.

Strategies include the following.

#1 Be Bigger, Faster & Cheaper

At the top end of town, large data warehouses are getting larger.  Multi TB data warehouses are common PB & multi PB are at the leading edge.  But at the same time the response time requirements are getting smaller.

Horizontally partitioned, distributed, highly scalable database platforms are the only way to fulfill these requirements.  Doing this on traditional platforms (Oracle, SQL Server, Teradata, DB2) can be difficult and/or costly.  If you can make it simpler, while being scalable, faster & cheaper (cheaper licenses, less hardware, less difficult to deploy & manage) you’ll have a good story to tell.

Netezza, Greenplum, Vertica & Aster Data are examples in this group.

#2 Be Smaller, Faster & Cheaper

At the other end of the spectrum there are organizations that just want their report queries to run faster.  They may not want to build a multi PB data warehouse, they may just have a few hundred GB of data and want snappy, easier report style queries to run quickly.  Providing a simple, cheap database platform that is easy to implement and easy to migrate to means organizations can quickly start receiving bang for their buck.

Kickfire & Infobright are examples in this group.

#3 Be Specialized

Similar to what I spoke about in the Enterprise OLTP post, picking a specialization and focusing is always a good method for getting a foot inside the door (albeit in a limited initial capacity).  Focusing on a specialization and packaging up your database platform with re-built tools applicable for that specialization can be a great way to out maneuver the competitors.  Such examples of pre-packaged tools include reports, dashboards, alters and other analysis targeted towards that specialization.  This can further save your customers time and money from removing the need to build such capabilities be-spoke.

Tenbase & SenSage are examples in this group.