JumalaSika ltd - Fotolia
Hewlett Packard Enterprise this week scooped up a dying asset to bolster its storage for AI and containers.
The vendor said it acquired MapR Technologies, a once-promising data management startup that misfired in its attempt to take on Cloudera and Hortonworks -- two leading Apache Hadoop providers that recently merged.
HPE did not disclose the purchase price for the MapR storage technology, but reports of MapR's demise have circulated for weeks.
It is the second big data acquisition for HPE since November, when it bought big data infrastructure startup BlueData. Customers can use BlueData to launch containers, processing or analytics as a service.
MapR storage software incorporates data management with analytics and machine learning. The MapR Converged Data Platform is a POSIX-compliant file system designed to enable containers, files and streaming data to be shared between clouds and data centers. The MapR name stems from Apache Hadoop MapReduce processing.
Mike Matchett, founder of storage consulting firm Small World Big Data, said MapR focused on the wrong segment of the market. He pointed out that while MapR included storage technology, it didn't position itself as a storage company.
"They never really wanted to sell to their technology strength, which was software-defined storage for big data. When I look at their product, I see a great storage layer that could have provided the next-generation data center platform. But they didn't want to go to market as a data center platform. They wanted to go to market as a data science platform for big data analytics," Matchett said.
Mapping MapR to HPE Synergy, GreenLake composable
Matchett said HPE's challenge will be integrating the technologies as it plays catch up to big data rivals like IBM and Microsoft.
"HPE has stayed the infrastructure course, but that market was getting disintermediated," he said. "They fumbled on Autonomy and a couple other [transactions] and knew they needed to build a healthier middle layer. With BlueData, they got some parts for having a containerized version of big data systems and a virtualization layer to all their enterprise storage. And then with MapR, they get an actual software-defined storage layer" for building large AI clusters at scale.
HPE has been trying to keep pace with competing storage vendors in fleshing out a hybrid cloud fabric that enables multiprotocol storage with automated management across physical and virtual domains.
Patrick Osborne, a vice president and general manager of big data and secondary storage at HPE, said MapR storage is used by hundreds of large global customers. He said HPE plans to integrate BlueData and MapR technologies to provide a scale-out storage layer for unstructured data.
"They have a pretty powerful software-defined platform," Osborne said of MapR. "Using MapR as a persistent store for AI and machine learning analytics, and also as a persistent data store for the containers themselves ... we're pretty excited about that. We also expect to use the platform in future use cases that are a bit more vertical-specific," including connected vehicles and autonomous oil drilling.
MapR's technology will also become part of HPE's Intelligent Data Platform and drive AI and machine learning applications.
Osborne said the MapR technology diversifies the options customers have with HPE Synergy composable storage infrastructure and HPE GreenLake hybrid cloud as a service. Osborne said about 200 MapR employees will join HPE.
MapR laid off 122 workers earlier this year and told California labor officials it was shutting down amid dwindling demand for Hadoop deployments. Investors poured in a reported $280 million of funding, including $110 million from Google. The company's last funding round came in 2017.