Data Warehousing and the DW Appliance

Today I attended a seminar presented by Oracle about the Exadata database machine. As usual it was presented as a compelling argument that a complete solution like a data warehouse appliance is one that ultimately reduces complexity and increases capacity.

During the session today, they quoted a study from TDWI which says that within 3 years 78% of respondents expect to be using a data warehouse appliance. I find this a staggering figure. Consider that today DW appliances have evolved and each of the major database vendors have their own flavour, but we still continue to see that traditional databases significantly outnumber appliances. I do admit that the argument for an appliance is one that needs to be considered. The idea that a single server or linked servers can come into a data centre and within days can be productive is incredible. The bigger question becomes which appliance do I choose? Do I buy Exadata, Teradata, Data Allegro, or Netezza? I guess that answer depends on what you want to achieve and the basic knowledge and comfort level you have with each vendor.

I find that the real differentiator is the functionality available within the appliance. As you might know Netezza uses Postgres as it’s base database. Although a great database, it is provided within only the basic SQL set of commands. To extend it out, you need to buy additional options and features. Then you look at a database like Oracle’s which has 30 years of development in it and now we are presented with a database that supports very advanced SQL. Oracle has provided an appliance that runs the database. It has not created a database which happens to run on an appliance. I often wonder how many people are aware of the advanced functions and analytics available in the Oracle database. The database has a significant number of statistical functions from the simple to the esoteric. The database contains a number of data mining functions as well. Although Oracle sells a great front-end tool in Oracle Data Mining (ODM), the functions which the tools use are available to all database users. So when you start to look at what appliance, you need to look at more than simple administration and technology, but at capability. The appliance who can do the most without any additional costs seems like a solid option. Now as to whether or not we will see a big increase  in the use of appliances will depend on how well and how cost effectively these appliances can replace our current hardware platforms.