SiSense gained a lot of traction last week at the Strata conference vr_ngbi_br_importance_of_bi_technology_considerationsin San Jose as it broke records in the so-called 10x10x10 Challenge – analyzing 10 terabytes of data in 10 seconds on a $10,000 commodity machine – and earned the company’s Prism product the Audience Choice Award. The Israel-based company, founded in 2005, has venture capital backing and is currently running at a profit with customers in more than 50 countries and marquee customers such as Target and Merck. Prism, its primary product, provides the entire business analytics stack, from ETL capabilities through data analysis and visualization. From the demonstrations I’ve seen, the toolset appears relatively user-friendly, which is important because customers say usability is the top criterion in 63 percent of organizations according to our next-generation business intelligence.

Prism comprises three main components: ElastiCube Manager, Prism BI Studio and Prism Web. ElastiCube Manager provides a visual environment with which users can ingest data from multiple sources, define relationship between data and do transforms via SQL functions. Prism Studio provides a visual environment that lets customers build visual dashboards that link data and provide users with dynamic charts and interactivity. Finally, Prism Web provides web-based functionality for interacting, sharing dashboards and managing user profiles and accounts.

At the heart of Prism is the ElasticCube technology, which can query large volumes of data quickly. ElasticCube uses a columnar approach, which allows for fast compression. With SiSense, queries are optimized by the CPU itself. That is, the system decides in an ad hoc manner the most efficient way to use both disk and memory. Most other approaches on the market lean either to a pure-play in-memory system or toward a columnar approach.

The company’s approach to big data analytics reveals the chasm that exists in big data analytics understanding between descriptive analytics and more advanced analytics such as we see with R, SAS and SPSS. When SiSense speaks of big data analytics, it is speaking of the ability to consume and explore very large data sets without predefining schemas. By doing away with schemas, the software does away with the  need for a statistician, a data mining engineer or an IT person for that matter. Instead, organizations need analysts with a good understanding of the business, the data sources with which they are working and the basic characteristics of those data sets. SiSense does not do sophisticated predictive modeling or data mining, but rather root-cause and contextual analysis across diverse and potentially very large data sets.

SiSense today has a relatively small footprint, and is facing an uphill battle against entrenched BI and analytics players for enterprise deployments but it’s easy to download and try approach will help it get traction with analysts who are less loyal to the BI that IT departments have purchased. SiSense Vice President of Marketing Bruno Aziza, formerly with Microsoft’s Business Intelligence group, and CEO Amit Bendov have been in the industry for a fair amount of time and understand this challenge. Their platform’s road into the organization is more likely through business groups rather than IT. For this reason, SiSense’s competitors are Tableau and QlikView on the discovery side and products likes SAP HANA and Actuate’s BIRT Analytics in-memory plus columnar approaches are likely the closest competitors in terms of the technological approach to accessing and visualizing large data sets. This ability to access large data sets in a timely manner without the need for data scientists can help overcome the top challenges BI users have in the areas of staffing and real-time access, which we uncovered in our recent business technology innovation research.

vr_bigdata_obstacles_to_big_data_analyticsSiSense has impressive technology, and is getting some good traction. It bears consideration by departmental and mid-market organizations that need to perform analytics across growing volumes of data without the need for an IT department to support their needs.

Regards,

Tony Cosentino
VP and Research Director