The Path to the PDW
For decades we had the traditional electric oven to cook with and people thought it was great. This technology allowed for consistent and dependable cooking for its users. Then something magical happened, the microwave was invented and used for food preparation. By leveraging the a different method to cook, people could make similar meals in a fraction of the time. This paradigm is where we currently are with data warehouse appliances. For years, we all leveraged our RDBMS and OLAP technologies and could depend on the results within reasonable response times. There were always a few tweaks we could apply to indexes, memory, or ETL processes to help with query performance on those stubborn long running queries. Now with the amount of data we can store, often these time honoured traditional approaches do not always work. Enter the SQL Server PDW – Parallel Data Warehouse Appliance solution.
The PDW is an all in one appliance that ships to you preconfigured and ready to use on day 1. It requires minimal configuration to connect to your network and start feeding it data. The PDW requires at least 1 base rack and 2 compute nodes. The appliance scales by adding additional nodes to the rack and then when a rack is full by adding expansion racks. You can see in the image below originally posted in Microsoft SQL Server 2012 Parallel Data Warehouse, how the Control node and Compute nodes are configured.
Query performance can be over 50 times faster than traditional systems, a la the Microwave oven. In addition to speed, the current PDW system can accommodate over 6 petabytes of data. For those of you leveraging Big Data solutions like Hadoop. The PDW can connect to those data sources as query it via Polybase technology. This enables the PDW to be the workhorse solution for all of your query and Business Intelligence data repository needs.