Back to Research

University of Toronto  

Project 1153 - Reducing JVM memory costs in the cloud

SHARE THIS POST

Running from 2022 to present

Reducing JVM memory costs in the cloud

Products
Open J9  

Managed language runtimes such as the JVM (Java virtual machine) provide many benefits, such as platform independence, ease of programming, and better security. Unfortunately, these benefits come at the cost of increased memory requirements, which is particularly problematic in cloud environments where the larger memory footprint results in lower consolidation ratios and reduced efficiency as fewer applications can be packed into the same hardware.

We explore two approaches for reducing the memory costs of running JVMs in the cloud, which we plan to implement in the OpenJ9 JVM and evaluate using representative cloud workloads.
- Leverage modern non-volatile memory (NVM) hardware such as Intel Optane DIMMs to reduce the monetary cost of memory. We research techniques for identifying objects that can be efficiently offloaded to persistent memory.
- Reduce the JVM memory footprint by unifying the management of Java heap memory and the memory used for just-in-time (JIT) compilation.

Learn More about the Research Team.  

Explore the product that harvests this research results  

Research team:

  • IBM Project Lead (RCL): Marius Pirvu, IBM
  • IBM Contributor (RCC): Vijay Sundaresan, IBM
  • IBM Contributor (RCC): Aleksandar Micic, IBM

Institution:

University of Toronto   

SHARE THIS POST