Richard Hackathorn: July 2010 Archives

Aster Data Systems is presenting with Sharmila Shahani-Mulligan, EVP of Marketing, Steve Wooledge, Sr. Director of Corp Mtg, and Stephanie McReynolds, Director of Product Marketing. Their slogan is "More Data and Big Insight" to enable ultra-fast, advanced analytics on big data volumes.  Aster Data was at BBBT in January 2009 as summarized with a blog here.

Aster Data currently has 110 employees and around 40 customers. Their annual revenue has doubled over last two years. They have scaled up their staffing across the company with an emphasis in sales, marketing and management.

Customers were initially digital media and Internal-based firms, but now represent financial services, national retailers, telecomm, with companies like Barnes & Noble, LinkedIn, and Intuit. Many customers have Aster Data servers that co-exist with Teradata, Oracle and IBM DB2.

In one customer example, Full-Tilt Poker, a leading online poker site, is using Aster Data for fraud detection. Reports generation was reduced from once per week to just 15 minutes. Further, the fraud detection can check 140,000 electronic poker hands per second and can react to fraud situations within 90 seconds.

Partners are SAS with joint development for in-DB analytics and go-to-market program, Carahsoft in federal sector, Dell with cloud services and PowerEdge C-Series hardware, Microstrategy, Terremark, and Informatica. Dell will provide full integration of Aster systems on PowerEdge C servers, doing the assembly, test, and delivery as a Dell order. Current competitors tend to be Vertica on the east coast and Greenplum on the west coast.

Sharmila remarked that 70% of the big data problem relates to data that lives outside of the primary data warehouse. The current use case focus upon understanding customers, decide/act situations, and complex systems monitoring.

We had a deep discussion on evolving definition for a data warehouse and the unique capabilities of MapReduce. The principle is to run analytics as close to the data as possible, eliminating data movement. Sharmila defined Advanced Analytics as iterative analytics that start with a few dimensions and expand to more dimensions. It should be easy to explore the data. Advanced analysis implies a more complex processing involving the reading data from the database, doing the analysis, and then writing the results back to the database.

Aster Data is now offering an Eclipse-based visual development studio plus a large library of functions for common analytics. We ended with an NDA discussion of their future product roadmap.

My Take...

Aster Data is gaining traction among early technology adopters. However, broader acceptance is limited by general industry acceptance of database paradigm shift. It seems that many BI professionals are sitting on the sideline waiting for others to mature and prove MapReduce technology. However, their current customers are impressive and number in the thirties. These customers are implementing innovative applications that enhance key business strategies, with use cases that are more varied. Aster Data's future products plans intend to enable further analytic innovation. My advice is: Watch this company, and understand their technology.

Infobright with Don DeLoach, CEO, Bob Zurek, CTO and VP Product Mgt, and Susan Davis, VP Marketing is presenting their open-source analytic database.

Infobright was founded in 2006 with headquarters in Toronto and offices is Chicago (for sales) and Warsaw (for development). Product introduction was in late 2007 with 120 customers with enterprise edition and 40,000 downloads of community edition (called ICE). Usage is gauged as activity in the community forums, which has 8,927 registered users (as of today) of whom are 4,000 are active users.

Don started with thoughts about recently joining Infobright in May. The company positions itself as "a high performance analytic database that delivers fast query performance against large volumes of data with minimal IT effort". Later, Don emphasized "simple, fast, low cost with small footprint".

As a customer of Infobright, Tim Moss, Chief Data Officer, from Bango, leader in mobile billing and analytics services utilizing a SaaS model. Large client generates 150M rows/mo or 450 GB/mo. When compared with Microsoft SQL Server, Infobright provided several orders-of-magnitude performance increase for queries. The data was compressed, in one example, from 450 GB to 10 GB for a month of data.

Susan shared the use cases for their 120 customers of the enterprise edition, as shown in the figure. Web analytics and data marts are the dominate use cases. Competition to Infobright tends to be: Vertica on the commercial side and MySQL with MyISAM on the open-source side

Bob went into the Infobright technology, which is based on Rough Set mathematics. Their competitive advantage comes from deep intelligence within the data. There are product gaps with: complex string, blobs, replication, some SQL syntax, multi-node, monitoring/managing tools.

Infobright is architected into MySQL, having an OEM agreement with Oracle to package MySQL into their downloads, as shown in the figure. Unlike the usual storage engine within MySQL, Infobright also replaces the query optimizer and includes a separate data loader. For more details, there is an informative 18-page technology whitepaper that is available as part of the ICE Documentation Pack. Also there is a section listing whitepapers on the website.

Bob ended on future directions and product roadmap. Some exciting developments are planned over the coming year! 

My Take...

Infobright's positioning is based on: low cost, performance, scalability, lower effort - which is not bad but assumes that potential customers have existing problems of cost and performance. I feel that they need to move up the food chain in information value by focusing on the business value of 'analytics'. They have packages in which Infobright is combined with Pentho, 

The internals for Infobright reminds me of fractal compression that squeezes all redundancy from the data by deducing all the rules buried within the data. In other words, they "pre-analyze" the data. If this observation is valid, then Infobright's strategy is backwards. Instead of focusing on performance, they should focus on deep data analysis. I am willing to pay $100 to double my query performance, but I am willing to pay $1,000 if you tell me what my data really tells me of my business. I wonder what an analysis/visualization tool that surfaces the information embedded within the knowledge grid/nodes.