A native of the Bluegrass state, Ken Gardner, president and CEO of Sagent Technology, was born, raised, attended college and held his first job as a computer programmer in Louisville, Kentucky. Graduating from college as a finance major, Gardner hoped his first job would somehow relate to stock market analysis. Unfortunately it was during President Nixon's 1972 wage freeze, and there were not a lot of jobs available. "I remember interviewing at a bank that had two positions available branch manager trainee and programmer trainee. I wasn't sure I wanted to be a branch manager trainee, but I knew I didn't want to be a programmer." One doesn't always get what one hopes for, as Gardner relates, "They offered me the programmer position! That's how I began my career in the information technology field. Three weeks into my training, I had a program running. A year later I had 10 people working for me." He says, "Career counselors are always advising kids to find their bliss. Well, I found mine. From that point on, computing basically became my vocation and my avocation."
Recapping the experiences that led him to form Sagent, Gardner says, "After working at banks and insurance companies as a CICS assembly language programmer, I was fortunate to be hired by a computer services company called Tymshare Inc. and worked in research and development for that firm. I left that firm to become vice president of technology at a human resources mainframe software company called Tesseract when it was a 40-person start-up." Gardner held a senior management position at Viewpoint Systems and was the president, CEO and one of the founders of ReportSmith which was later purchased by Borland International. When Gardner left Borland, he and John Zicker, who is Sagent's executive vice president of technology, founded Sagent.
Today Sagent's charter is to bring information closer to decision-makers in a way that is quicker and easier to use. Sagent seeks to do this with the Sagent Data Mart Solution, an integrated package of best-of-breed products, coupled with Sagent's comprehensive professional services.
Sagent's success will hinge on certain early choices made by the founders. Gardner says, "We made a number of pretty significant technological and architectural bets back in 1995 when we started the company. One of those bets was choosing NT as our platform." Commenting on that selection, he adds, "In June of 1995, we essentially made a bet that Microsoft in conjunction with the hardware vendors would move the NT platform forward so that it would become an acceptable platform for large-scale data warehouses. And although they can't quite get to the top 5-10 percent of data warehousing applications, with SQL Server 7.0 on an eight-way server running NT 5 they will get pretty close to touching the top."
The other strategic moves made by the Sagent co-founders involved developing the data flow approach, providing an end-to-end solution and focusing on performance and scalability. "Those were really big bets, and for a long time the analyst community consistently told us that they didn't want to write about Sagent because we did not fit into their categorization scheme. Our product offering did not fit neatly into any single category since we were providing an integrated solution. While strategy bets like that don't always pay off, they certainly did in our case. Now the marketplace is moving toward total solution approaches," Gardner states, "and everybody is attempting to put together the pieces that we already have in place."
According to Gardner, continuing advances in storage technology have made decision support systems much more affordable. "I remember when we first started Sagent one of the reporters asked me to predict data mart size. I responded that we were figuring about five gigabytes. Today we're doing projects with telcos where a data mart of 600-700 gigabytes is not unusual," explains Gardner. "All of a sudden the idea that you can keep a couple of terabytes of history around is not a crazy idea. Not very long ago, it was truly a crazy idea. In the late eighties, storage cost about $100 per megabyte. Today it's only about 10 cents a megabyte. The fact is that all of the technology advances have made decision support systems very economical."
In order for a decision support system to be successful, rapid and reliable performance is a key requirement. "From the beginning we identified that most of the performance issues in these systems are architectural issues," Gardner says. "We basically decided that the only way to positively affect performance in the business intelligence/data warehousing space was to build architectural solutions to those problems. At the most basic level it is not complicated. It really comes down to four things that one has to manipulate: CPU cycles, I/Os, memory and bandwidth."
To explain how Sagent addresses these problems, Gardner uses an example of a retailer with a data center connected to the corporate offices by a T1 line. "With 400 users, 350 stores and about 1,000 transactions per day per store, the retailer realized the benefit of being able to determine the top 10 and bottom 10 product sales by store, by region and at a consolidated company level every day against 350,000 rows. However, when the 400 users began work at 8 a.m., they all launched the same query. In a two-tier environment, that ends up being about a billion rows attempting to move across that T1 line. If they were lucky, everyone would have their answer in four hours! Certainly not acceptable. With the Sagent Data Mart Solution, that retailer would experience an improvement in response time. Our Data Mart Server is capable of doing ranking in the application server. So instead of moving a billion rows, you move 8,000 rows in a matter of seconds versus hours. That's a really simple example of how architecture affects performance," Gardner explains.
Looking ahead, Gardner sees rapid performance as a continuing requirement for success. "The biggest opportunity moving forward for everybody in the business intelligence space is information systems that are attached to the Web. People are demanding 'snap-of-the-finger' response, and they don't tolerate anything less. The revolution that has occurred in the Web is the 'stop' button. How long do any of us wait for anything? We think it is about 30 seconds. If the user gets response within 30 seconds, it's fine. If not, the user will hit the 'stop' button and go to another Web site."
Continuing, Gardner says, "There are four architectural aspects that are required to maintain a high-performing Web-based decision support system. The first involves the back-office systems which are very complex. In some cases, the word 'legacy' would be a very kind word. From these systems, you have to be able to create a single business view." Gardner says getting this "single" view is a challenge for many companies. He explains, "The average organization is running somewhere between eight and 15 back-office systems. When a customer comes to an organization's Web site, the customer expects the organization to know everything about any and all past transactions. What this requires and this is dramatically different from what IT has had to deal with in the past is the ability to give the customer the current and complete status of his or her account. That information must be compiled from all of the back-office systems. And, I can tell you that this is a big problem for a lot of organizations. In general, transaction systems that have been put in place in the back offices years ago are still running on mainframes, on AS400s, on Wangs and on just about anything you can name. Organizations have these very complex back-end environments, and they have to be able to source data out of those environments and create the single view," Gardner states. "We think that should be accomplished with a star schema, and our Data Flow technology includes components that allow you to automate the creation of that star in order to obtain that single view," he emphasizes.
Pioneered by Sagent, data flow technology streamlines and simplifies the data mart process. Gardner explains, "With our Data Flow architecture, we take a process flow approach to loading the data. We don't take a mapping approach. There are things that mapping approaches can do really well, but one of the things they don't do well is handle a lot of complexity. Our Data Flow technology provides a visual programming environment to import data from multiple sources, perform complex transformations and load data into multiple targets."
"Getting fast performance from the database," Gardner says, " is the second consideration. In the relational OLAP marketplace, there are two ideas that compete. One is the star and the other is the snowflake. We think the star schema is the most effective data model for supporting the dynamic, high-performance query and analysis demands of today's corporate decision-makers, and we have technology built into our data access server that understands stars and can exploit them."
With the data loaded and reorganized, the next consideration is scalability to hundreds and thousands of users. "We are continuing to architect our product to achieve the scale needed to support large organizations. Our goal is to be able to support 100 concurrent users or 1,000-2,000 total users. If you have more than 2,000 users, we can run multiple data access servers synchronized by a single repository," Gardner emphasizes.
"The last piece has to do with handling the 'last-mile' problem delivering the data to the end users," Gardner says. "Our solution automatically utilizes aggregate navigation to deliver superior query performance."
"We now have a very broad product line because we have addressed the four critical areas. You have to be able to create the single view, you have to be able to exploit it, you have to be able to scale, and you have to handle the 'last-mile' delivery part of the problem if you're really going to be a solution vendor. We believe that we have accomplished that with the Sagent Data Mart Solution."
Sagent recently added a professional services division to assist customers with data mart implementation. "With Sagent Professional Services, we develop a knowledge transfer environment with our customers. We'll build the first star while the customer watches, and we'll build the second one together. On the third star, the customer will build it and we'll watch. I want to emphasize that we're not selling data marts in a box. We engage in a relationship with our customers and train them to be successful," Gardner emphasizes.
Ever since Gardner at age six beat his father playing chess, using strategy to succeed has motivated him. Gardner's skill at anticipating the market requirements and developing a product line to address those requirements has enabled Sagent to grow from a small start-up to its present position as a leading provider of turnkey data marts.
|Sagent Product Components|
|Key components of the Sagent Data Mart Product Solution product suite include:|
|Sagent Data Load Server. Sagent Data Load Server extracts data from multiple client/server and mainframe databases, transforms that data into a star schema data structure and then loads the data into a Sagent data mart.|
|Sagent Data Access Server. Sagent Data Access Server delivers the data loaded into a Sagent data mart to end users. The server is designed to allow large numbers of users to access and analyze data stored in star schema structures.|
|Sagent WebLink Server. Sagent WebLink Server is a high performance, scalable application server that delivers information from the Sagent Data Access Server, allowing end users to query, analyze and report business information from a Web browser. The server also provides management capabilities that maintain the security and availability of Internet connections.|
|Sagent Information Studio. Sagent Information Studio enables users to access and analyze an organization's information in client/server environments. Information Studio, as well as WebLink Server, can be integrated with Microsoft Excel to aid in exporting result sets to spreadsheets for further analysis.|
|Sagent Design Studio. Sagent Design Studio provides a visual environment for describing data and designing the flow of data for both loading and accessing a data mart. Sagent Design Studio minimizes the requirement that users have in-depth knowledge of databases, networks and operating systems and allows them to concentrate on the business purpose of accessing, analyzing and delivering information.|
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access