The reasons, hurdles, and benefits
SQL Server 2008, with its built-in SQL Server Analysis Services (SSAS), SQL Server Integration Services (SSIS), and SQL Server Reporting Services (SSRS) business intelligence (BI) subsystems, is a recognized leader in the enterprise BI space. In this interview, SQL Server Magazine’s Megan Keller and Michael Otey talk with IIan Wajsman, director of IT Data Systems at Stein Mart, and Fausto Ibarra, senior director of Microsoft Business Intelligence, about Stein Mart’s experiences migrating from a legacy platform with limited BI reporting tools to SQL Server 2008. IIan shares some of the main reasons why Stein Mart made the move to the SQL Server platform, the hurdles faced along the way, and the benefits that Stein Mart attained from the move.
Ibarra: As you know, most companies today are facing a data explosion with terabytes or petabytes of data that they need to manage. Forrester research estimates that the amount of data in large databases is growing at 65 percent per year. One company facing such a growth in data is Stein Mart. Stein Mart is a large department store chain that’s based in Florida. And, like so many other retailers, it’s always looking for ways to get better intelligence and better insights into its business. Specifically, it was looking for a new BI system that would offer faster data reporting and new ways to analyze data. It had previously been using a BI system based on technology by IBM. Stein Mart chose a data warehousing and BI solution built on SQL Server 2008 with hardware from HP. As a result of this migration to a new system, Stein Mart not only got new tools for business analysis and forecasting but also was able to reduce the time it takes to generate its weekly sales reports from 14 hours to only 3 hours. And it was able to get significant savings of the order of $600,000 per year in IT operation costs. So, let me introduce you to Ilan, who’s going to be walking you through the project and what Stein Mart found out.
Wajsman: At Stein Mart, we faced some technical limitations because of our infrastructure and we needed to increase our BI capabilities. The existing solution we had was running on IBM AS/400, with an IBM DB2 data warehouse. Basically, we were reaching the limits in terms of table size and being able to scale out. Fausto talked earlier about data growth. Our data warehouse has actually doubled in size over the last six months.
SQL Server Magazine: We weren’t aware that there were any limitations on data size in the AS/400.
Wajsman: The version we were running was V5R4. There were not only limitations on table sizes, but we also weren’t able to partition the data. In the book, it says you can do partitioning, but it never worked. Plus, all of our extraction, transformation, and loading (ETL) code was custom written, so integrating with systems outside the AS/400 was a challenge.
SQL Server Magazine: Which BI platform did you use?
Wajsman: The back end itself was DB2. For reporting, we used MicroStrategy. So, as we were looking at different solutions and the costs of maintaining the system, we saw that there was considerable opportunity for not only modernizing our capabilities but also reducing our overall costs.
SQL Server Magazine: When did the actual migration take place?
Wajsman: We went into full-time production in June 2010, and it was roughly a nine-month process. Most of the migration time was actually spent migrating our ETL code from RPG and COBOL to SSIS packages.
SQL Server Magazine: Which of the SQL Server BI solutions did you implement?
Wajsman: We're using SQL Server 2008 for our data warehouse, and all of our ETL code is in SSIS packages. For the front end, we’re using SharePoint and SSRS for a lot of our reports. We're using PerformancePoint dashboards, and using Report Builder to build some of the reports. And we’re still using MicroStrategy for some of the legacy reports. We’re using SQL Server 2008 R2 for our analysis server.
SQL Server Magazine: What is the size of your data warehouse?
Wajsman: Uncompressed it was seven terabytes. When we moved it into SQL Server, it compressed down to two terabytes. Currently, it's a little over four terabytes and growing.
So, there are two main reasons we moved to SQL Server. One is its BI capabilities, and the other is the cost. Fausto mentioned that we saved $600,000 by moving to SQL Server. In our old environment, we paid about $50,000 a month in maintenance costs for licenses for the OS and the hardware itself.
SQL Server Magazine: Wow, that's expensive.
Wajsman: Yeah, and that was just for the hardware and the licensing, not to mention the amount of labor to keep that system running. So, we had quite a lot of options in terms of being able to apply that money to get a full-blown system without going beyond our existing budget the first year.
