The announcement of Cloudera acquiring Hortonworks has created uncertainty and confusion for their customers due to significant overlap between the two companies’ offerings. Redundant products will need to be “rationalized” or eliminated, causing disruption for their customers. The merger is focused on cost efficiencies in the sales force and support model. It creates a negative technology situation, forcing customers to shift, move, or make choices.
While Cloudera/Hortonworks announced they will ship the “Unity Release” eventually, every customer will be shown the path to move to a common platform because the merged company will be unable to sustain two platforms.
MapR identified a minimum of six areas of overlap that need to be rationalized:
In addition, the two companies have no clear offerings in four areas critical to customers today:
AI/ML — Neither Cloudera nor Hortonworks have a good AI/ML story because their platforms do not support POSIX and AI/ML libraries like MapR does. The MapR Data Platform was built in anticipation of the evolutions of workloads from Hadoop to Spark to AI/ML. Unlike the competition, MapR supports these workloads on a single platform and on one cluster in production.
Hybrid and Multi-Cloud — Neither Cloudera nor Hortonworks support seamless environments that span on-premises deployments to one or more clouds. MapR today supports hybrid and multi-cloud environments with open APIs, letting you avoid cloud lock-in and continue to run legacy applications as is. Additionally, MapR helps you lower cloud costs by up to 30%.
Containers — Neither Cloudera nor Hortonworks support containerized stateful applications. MapR is a leader and early adopter of both Docker and Kubernetes, and has made containerized stateful applications a reality. With the MapR Data Platform, data from containerized applications is persistent, scales as containers grow, and is protected with replication, mirroring, and instant snapshots.
Operational Analytics — Neither Cloudera nor Hortonworks support operational analytics. MapR lets you apply analytics in real time to operational applications. Only MapR supports mission-critical business applications under production SLAs, all without compromising on data consistency.
While Cloudera and Hortonworks continue to “rationalize,” MapR will be using this time to further innovate and expand our offerings. MapR is dedicated to delivering the capabilities customers want now for greater agility through containerization, seamless hybrid and multi-cloud, and a built-in path to AI/ML from an analytics environment.
Additionally, MapR delivered a number of enterprise-grade capabilities in 2011, including: high-availability, disaster recovery, NFS, and a single security model. Cloudera and Hortonworks have yet to deliver on these capabilities. Given Cloudera’s and Hortonworks’ underlying architecture, they are unlikely to get there anytime soon.
|MapR Clarity||Merger Mess|
|Data Science||Open APIs to enable ML and AI on the same cluster.||Redundant Offerings:
Data Science Workbench vs. IBM DSX
Siloed clusters, data movement required. No support for Python ML.
|SQL||Most open platform for SQL - Hive on MR, Hive on Tez, Spark, Drill, Impala. End-to-end JSON flexibility/schema-less queries.||Redundant Offerings:
Impala vs. Hive, Hive on Spark vs. Hive on Tez
No support for data exploration and operational analytics
|Security||Unified, common security model at platform level. Ubiquitous data protection, expressive and granular row/column access controls, and built-in auditing.||Redundant Offerings:
Sentry vs. Ranger
Add-on tools for security means inconsistent security models and access is not addressed at platform level.
|Management||Unified tooling for all data. Advanced management features with multi-tenancy, quotas, data placement, snapshots, and mirroring. Built-in High Availability.||Redundant Offerings:
Manager vs. Ambari
Hard to manage all data, even with many tools
|Governance||Enterprise-wide data catalog||Redundant Offerings:
Navigator vs. Atlas
Incomplete governance. Downstream dependencies on disparate security tools.
|Hybrid and Multi-Cloud||Built for hybrid/multi-cloud. Transparent data synchronization across on-premises and clouds, global namespace, built-in streaming, and storage optimizations to balance performance, capacity, and cost.||Redundant Offerings:
SDX vs. DataPlane
Cannot support multi-cloud natively, no global namespace. Concerns with cloud data management and consistency.