Par Stream has raised $5.6 million for further extending its big data analysis technology that uses in-memory and massive parallel processing to do real-time analysis.
The Series A financing was led by Khosla Ventures with Baker Capital, Crunch Fund, Data Collective, Tola Capital and private individuals participating in the round.
ParStream, which was founded in 2008, was previously Germany-based but now keeps offices in Cologne and Palo Alto. The search technology compresses data using bit map indexes. A bitmap index data structure allows for data to be read very fast as it resides in-memory. The data is stored in columns, which is better for analytical purposes, and focuses on fast response times and parallel processing. The company’s targeted use cases are customers that have huge amounts of data that is imported continuously just once and not changed afterwards. Web analytics, fraud prevention, online advertising, telco billing, smart metering and sensor network data analysis are all examples of potential use cases.
ParStream is a counter of sorts to other big data analytics providers. Interestingly, it does not use Hadoop, the distributed analytics file system. CEO Joerg Bienert said the data requirements are just too much with Hadoop. “It’s not great for real-time analytics,” Bienert said.
Instead, due to the compression it does, ParStream analyzes smaller pieces of data to get the analysis in real-time.
Par Stream is a likely acquisition target as it competes with HP’s Vertica and EMC’s Greenplum, both of which were once independent companies.