Briefing Notes: SiSense, the In-Chip Data Analytics Company

0 met with SiSense - the In-Chip Big Data analytics company at Structure 2013. Bruno Aziza (VP, Marketing) and Eldad Farkash (CTO) demoed their flagship Prism BI software and explained the technology behind it. Here is the summary.


SiSense has offices in San Francisco and New York and its R&D is in Israel. They are led by an impressive team of strong business and technological leaders. They were recently in news for their massive funding round of $10M. For the past few years, they have been strengthening their core technology while continuing to acquire more and more customers, including the Department of Fertilizers, Government of India. At Structure 2013, they announced their partnership with Rackspace to host their In-Chip Analytics solution on Rackspace Private Cloud.

Prism 10X and ElastiCube

 SiSense's Prism 10X is the first In-Chip BI Software that is capable of analyzing terabytes of data in seconds without the need for expensive cluster like hardware. It can run on commodity hardware, public cloud or private cloud. It

 is built on ElastiCube technology which is again based on vectorization techniques. Data from any structural source - SQL, Excel, Google Spreadsheet, Hiveor Salesforce is parsed in memory, translated in to columnar data and processed in cache. ElastiCube's in-memory querying loads and unloads from RAM on-demand, which means the in-column data that is being processed is in L1 cache, thereby making computations extremely fast. As part of processing, it can automatically derive relationships between columns, can suggest correlations across multiple data sources if needed. For example, if you have the employee information in your SQL database and sales information in SalesForce, you can import from both the sources, and Prism would be able to correlate Sales Metrics from SalesForce and the Employee ID from SQL table.

 ElastiCube engine also keeps performance usage patterns in memory. Imagine something like remembering math tables, say 10 *10 = 100, so that next time you see 10 * 10, you automatically write 100, instead of doing the math. This also enables incremental processing of data analysis. Say in the above example, you queried for some employee details after loading the Employee table from SQL database. When you load the Salesorce sales data next, SiSense still keeps queries on Employee table in memory, so that you don’t need to reload the SQL table.

 Pretty cool, right? See a live demo here.

 SiSense were able to win the Big Data Startup competition during Strata 2013 using this technology over other Hadoop based solutions. This blog post explains how they were able to analyze 10TB of data using a $10,000 box in 10 seconds. For more in-depth details, request for their whitepaper here.


Please check out their tiered pricing plans here. Check out their full software as a free trial download here.


 SiSense has been silently acquiring customers all over the world - Target, Samsung, Dept. of Fertilizers to name a few. Last year they grew over 520% in revenue. With fresh cash in hand, they are planning to double their team in size by the end of this year!

 CloudStory Summary

 SiSense has the right combination of technology, team, domain and solution. SiSense is here to disrupt the Big Data Analytics space and rock the Big Data world!

 With such a cool technology, opportunities are endless - we see specialized Big Data solutions running on a commodity box using SiSense solutions. For instance, an appliance to crunch Genome data built on commodity hardware powered by Prism. If you are interested in building such solutions, please reach out to Bruno on partnering as an ISV.


Updates from around the world