Over more than a decade SAS Institute, a US-based vendor of BI solutions, is a top player in this field. SAS’ roots are at North-Carolina State University and since 1976 the company is managed by its inspiring chairman Dr James (Jim) Goodnight. Today SAS is still a privately held company with an annual turnover of more than 3 billion dollars.
A couple of years ago SAS started profiling itself as a vendor of Analytics solutions. Ever since its foundation the company had a strong preference for Data Mining and today it claims the Analytics market as its own. Although the main part of SAS’s turnover emanates from license sales of the SAS BI-platform – which includes ETL, data warehousing, data quality, reporting and analysis – the full focus is on Analytics.
During the annual SAS meeting for analysts – held in a small village near Milan, Italy – it was all about Analytics. 50 analysts from all over EMEA were invited to get a full update by the complete C-level staff of SAS.
“The more complex your problem is, the better friend SAS becomes,” said Norman Manley, Managing Partner of Passionned Group. “SAS offers a number of capabilities I find quite impressive, especially in comparison to other vendors. SAS can accomplish very difficult tasks.”
First, he mentions SAS’ applications for fraud detection. Manley knows more vendors tried to embed the needed intelligence in their applications. “Not very successfully, though, but SAS really can do it: today it offers over 300 applications for fraud detection and pattern recognition. Very impressive.” Manley considers the in-memory processing of any volume of data to be an important new development. In a demo, it took Dr Goodnight 3.5 seconds to perform a simple analysis on a data set. “That doesn’t sound special. But it is – if you know the data set consisted of 3.5 billion records.”
SAS has agreements with amongst others database vendors Teradata and Greenplum (EMC). That’s all in field of SQL, said Manley. “But SAS developed a mathematical model and built all kinds of interfaces on various levels. That makes it possible to run their analytic software on other types of databases, such as Hadoop. For truly difficult tasks SAS has a database of its own, an unstructured database, where users can enter all kinds of data in any amount. This machine, which was demonstrated during the analysts meeting, has 1,000 processors. The earlier mentioned 3.5 billion records are distributed in portions of 3.5 million records per processor. Special SAS software makes the processors work parallel. That’s a new idea: there’s no more structure. No more data warehouse designs with snowflakes and star schemes, or DataVaults; the whole discussion between Kimball and Inmon vanishes. The unsorted voluminous data simply go into the memory where they are analyzed by 1000 processes.”
SAS extended its software portfolio with a new family, which Manley simply called ‘data mining’. “Perhaps it’s not the right name, but it’s where it’s at: there is an enormous mountain of data and you ask the software whether it’s able to figure it out. Of course, not just like that, this is based on are all kinds of very complex models and scenarios that search for correlations and patterns. However, it’s the whole Goldrush-idea: filter all the data and see if a nugget remains. As an example: Verizon is looking for the right pricing. Today it’s possible to load all the data of the last 10 years in a database, combine it with dates of birth and postal codes, and ask the software if there are any patterns. Similar to what food retailer Albertsons did, but a billion times bigger.”
A number of those models are very complicated, so it used to take hours to run them. Today, SAS can do that real-time, which is very important for decision management, as shown by decision “guru” James Taylor recently at a congress. He states that a lion’s share of decisions – i.e. most of the wrong ones – are made at the workplace. Norman Manley understands. “The decision to grant loans to people who can’t afford it, is not made by board members, but in the lower levels of the organization. Today SAS is clever enough to trace this wrong decisions by looking very closely at every transaction in real time, so without any delay in the process. Moreover, these systems are so called ‘learning systems’ that improve models and get more sophisticated over time.”
SAS emphatically presents itself as a vendor of solutions for Analytics. “Unfortunately, some people still think that Analytics is only BI-with-lipstick,” Manley concluded. “But SAS over-demonstrates the added value of this relatively new domain. And all-in a structured, technologically high-level and scientific way.”
Additional and more up-to-date information on this BI & Analytics-platform is available in our 100% vendor independent Business Intelligence Tools Survey 2019.