SAS Institute, Sun Claim Top ETL Throughput

Published: May 25, 2006

by Timothy Prickett Morgan

With data warehouses getting bigger and bigger as companies put more historical data online and keep richer data sets with which to do their correlations, getting data out of production ERP, CRM, and SCM systems and into a data warehouse can be a big, time-consuming, pain in the neck. That is why speed on extract, transformation, and load (ETL) is so important.

Server maker Sun Microsystems and data analytics specialist SAS Institute last week announced a benchmark result on SAS' ETL software running on Sparc/Solaris platforms, which it claims is a world record. The benchmark yields about twice the ETL throughout of a setup the two vendors tested in March 2005. This time around, Sun configured a Sun Fire E25000 server with 48 dual-core UltraSparc-IV+ processors running at 1.5 GHz. This machine had Solaris 10 and the SAS Enterprise Data Integration Server on it, and it was able to chew threw 4.29 TB of raw data in 70 minutes, or about 3.68 TB per hour. Last March, Sun had a Sun Fire 15K with 72 dual-core UltraSparc-IV processors running Solaris 10, and this box could only eat 2.2 TB in 90 minutes, or about 1.5 TB per hour. This once again demonstrates that the UltraSparc-IV+ chip is what Sun should have had to market years ago, and it is a good thing that Sun has managed to get it to market in September 2005.