MapR, one of the three big vendors of the Hadoop open-source big data software, is today announcing MapR Streams, a new piece of software for sending many kinds of data around a company.
MapR Streams is a type of publish-subscribe messaging system. It’s comparable to a tool like Apache Kafka.
“Performance is basically very similar to Kafka,” MapR cofounder and chief executive John Schroeder told VentureBeat in an interview. (Benchmarks aren’t yet available; they will be when Streams becomes generally available early next year.)
But there are a few differences. Unlike Kafka, Streams doesn’t require a second cluster of servers alongside Hadoop. Oh, and unlike Kafka, Streams isn’t going to be available under an open-source license.
MapR has traditionally supported proprietary tools that other Hadoop distribution companies haven’t, including the MapR-DB database and the MapR-FS file system. So this strategy isn’t very surprising, even it does buck the trend of companies open-sourcing key underlying technologies, like Microsoft with ASP.NET (and more recently Chakra) and Apple with Swift.
With Streams, Schroeder points to advantages in terms of cost and complexity.
The integration of messaging infrastructure helps MapR compete with the likes of Hortonworks, which acquired Onyara earlier this year, as well as Intel-backed Cloudera, which recently announced the Kudu open-source storage engine.
Messaging is becoming more important lately, with startup Confluent taking on funding to commercialize Kafka, Microsoft pushing the Azure Event Hubs hosted service, and Amazon Web Services offering Kinesis Streams.
MapR started in 2009 and is based in San Jose, California. The company announced a $110 million funding round in June 2014 and now has 350 employees. Last month, the company said it had achieved a 160 percent year-over-year increase in bookings for the third quarter of 2015.
“We’re going to be doubling the sales force again next year,” Schroeder said.
There’s more on MapR Streams in a blog post.