Supercharge Big Data Analytics with Hive on MR3

Hive on MR3 is a high-performance alternative to Apache Hive and Spark SQL, running natively on Kubernetes.

What is Hive on MR3?

Hive on MR3 is a powerful, cost-effective, and portable way to run Apache Hive workloads. It combines the familiarity of Hive with the performance and efficiency of the MR3 execution engine.

Fast and Correct

Achieve high performance without sacrificing correctness.

Unified Processing

Run interactive and batch queries side by side in a single system.

Runs Anywhere

Run in any environment with flexible compute and storage options.

Consistently Fast and Always Correct

Hive on MR3 runs slightly slower than Trino for sequential queries, but significantly faster under concurrent workloads on the 10TB TPC-DS benchmark. Unlike Trino, it returns correct results for all queries.

With Hive on MR3, you don’t have to choose between performance and correctness.

Sum of running times, 10TB TPC-DS

Hive 4 on MR3

0 sec

Trino 468

0 sec

Spark 4.0.0-RC2

0 sec

Hive 4 on Tez

0 sec

10TB TPC-DS Benchmark

Check Out Details

Unified Batch and Interactive Processing

Hive on MR3 features fault-tolerant execution and built-in capacity scheduling. By leveraging capacity scheduling, interactive queries can be prioritized while batch jobs continue running reliably in the background — ensuring smooth operation within a single unified system.

Runs Anywhere with Compute-Storage Separation

HDFS

Unlike Apache Hive, which runs only on Hadoop, Hive on MR3 runs in any environment — on Hadoop, on Kubernetes, or even without a resource manager. It works with both HDFS and S3, enabling full separation of compute and storage. You can deploy it on-premises, in the cloud, or both.

Why Hive on MR3?

Hive on MR3 can run both interactive and batch queries together, simplifying operations and reducing costs. With fast autoscaling, smart caching, and easy deployment, it offers a powerful combination of performance, resource efficiency, and portability.

Operational Efficiency

Simplify operations and reduce costs with a single system for all workloads.

Resource Efficiency

Maximize resource efficiency with autoscaling and smart caching.

Deployment Efficiency

Set up fast with automation scripts and production-ready configurations.

Simplify Operations and Reduce Costs

Many organizations deploy separate systems for interactive and batch queries, increasing complexity and costs. Hive on MR3 streamlines operations by offering a single fault-tolerant system that handles both workloads. Capacity scheduling ensures efficient resource usage without compromising performance.

With Hive on MR3, one system is all you need.

$$$$

Dynamically Scale and Cache Smarter

Hive on MR3 features fast autoscaling in cloud environments, enabling efficient use of resources as workload demand changes. Its fault-tolerant design ensures reliable execution, even when spot instances are reclaimed. Smart caching, which can be enabled selectively, reduces traffic to data sources like S3, minimizing unnecessary access.

Deploy Hive on MR3 in 30 Minutes ⏱️

Hive on MR3 offers multiple deployment options: shell scripts for all environments, and Helm charts and a custom TypeScript generator for Kubernetes. With quick start guides and production-ready configurations, users can get Hive on MR3 running in 30 minutes — start to finish.

Upgrade Apache Hive 2 or 3 to Hive on MR3

Still running Apache Hive 2 or 3 on Hadoop? Hive on MR3 makes upgrading easy, with no need to overhaul your infrastructure. Choose the path that works best for your environment.

Keep Your Hadoop

Run on your current Hadoop cluster with a compatible build we provide.

Install Kubernetes

Deploy Hive on MR3 on Kubernetes alongside your existing Hadoop cluster.

Run Standalone

Use Hive on MR3 without Hadoop or Kubernetes. Just deploy and start.

Ready to Experience the Power of Hive on MR3?

Try Hive on MR3 today and see the difference.

Get Started Contact Us