May 17, 2018

2M euros for Indexima

Indexima, a french emerging player in BI/Big Data, just announced a new VC round of 2 millions euros. It will help the team to grow up to 15 around the end of the year.

The company, founded in March 2016 in Paris by Florent Voignier and Nicolas Korchia, addresses the query latency aligned with number of entries. The goal is to deliver a constant and very small response time whatever is the number of rows, tables and globally size of data warehouse. The idea came from what they accomplished at Mappy in 2014 where queries came from 20 to 0.1 seconds on top of a Hadoop cluster without any additional nodes. This approach then got selected by Natixis and Canal +.

So how does it work? Indexima, as its name invites us to think, is a new layer on indexation residing between the data warehouse and the visualization service. The team builds a Data Hub based on 3 pillars and with points taken from their web site:
  • HyperIndex are multi-dimensional distributed and persistent "in-memory" indexes that include pre-aggregations to avoid ALOP cubes.
  • DataSpace is a single secure access point to all indexed data. Algorithms learn from past requests to understand query trends and behaviors to build in real-time the right HyperIndex.
  • K-Store is a column-oriented file format, fully indexable, optimized for S3 and HDFS, the storage layer standard for the cloud and included in Hadoop. The company plans to offer this specification to the open source community. Important K-Store is not a database nor a file system.
The Data Hub supports various partners, key for their adoption:
  • Data Applications: Tableau, Qlik, MicroStrategy, Power BI, Excel, Talend, Dataiku, Jaspersoft, Cognos and SAP.
  • Data Sources: Hortonworks, Cloudera, MapR, Oracle, IBM DB2, Teradata, Amazon, Azure, SAS, SQL Server, Salesforce and ServiceNow.

Next steps for the company will be to consider and support Presto, Spark ou Hive but the company delivers already faster results then Spark. We expect the company to land also in US as this technology can solve big data issues there as well and help them to boost their business to finally take off.

Interesting approach, I'd like to see how they can compete against GPU database such MapD, Kinetica or more "classic" in-memory approaches such Aerospike or GridGain.

GDPR and data cataloging seem to be also a direction for Indexima and for the latter, competition will be more intense with Alation or Waterline Data to list a few players.

But again this fantastic story shows the difficulty France investors continue to demonstrate with a ridiculous financial round. Please consider the speed dimension especially in that space against USA or China competition.

0 commentaires: