The Oracle Customer Experience Industry Framework (CXIF) powers many of Oracle’s industry-specific offerings. This framework is built using Helidon and GraalVM Native Image. CXIF uses GraalVM Native Image to create minimum-size (< 50MB), precompiled executables of its microservices. They also emphasize the security aspect of AOT compilation with Native Image.
The Messaging team at Disney Streaming sends millions of messages to customers for use cases such as password recovery, account changes, and purchase confirmations. While researching ways to mitigate the cold starts of their serverless Java workloads, they discovered that a function that took 3.6 s to start on the JVM, started in under 100 ms once compiled using GraalVM Native Image.
Adyen, a payment processing platform, offers services such as credit cards, bank-based payments, as well as digital transactions. The company was looking for ways to modernize and run its C/C++ payment application in the cloud. They adopted and used GraalVM to migrate their existing terminal-based payment applications to the cloud. This also helped to mitigate their security concerns thanks to the sandbox environment that GraalVM provides.
In Oracle Database (21c version) developers can now run JavaScript code snippets inside the database, where the data resides. The Multilingual Engine (MLE) in Oracle Database 21c is powered by GraalVM: GraalVM Native Image compiles the MLE runtime and all required GraalVM components into a shared library that is loaded on-demand into a database process. See how to run the Oracle Database Multilingual Engine yourself.
Facebook, a heavily visited social media platform, uses Java in areas such as big data (Spark, Presto, etc.), backend services, and mobile. Facebook decided to evaluate GraalVM as a Java runtime. By just switching to GraalVM, without any code changes, they managed to accelerate Spark workloads around 10%-42% and reduce memory and CPU usage.
Twitter, one of the most visible social networks, runs ~1,000s of JVMs in multiple datacenters. The company was looking into ways to increase availability of the platform while keeping an eye on costs. Twitter saw GraalVM as an optimizing compiler and JVM, and tried it. Running the Tweet service on GraalVM, Twitter has achieved 8-11% CPU saving while requiring 18% fewer machines.
Standard Chartered Bank, an international banking and financial company, uses mostly Java for development and looked for a way to make Java applications start fast, adapt for cloud deployments, and streamline their CI/CD pipeline. To improve applications performance and make them cloud native, the engineering team considered using GraalVM Enterprise as a Java runtime.
NetSuite provides a set of cloud-based business management services encompassing ERP, Financials, CRM and e-commerce for more than 19,000 organizations. The engineering team, working on its next gen recommendation system, used GraalVM and grCUDA to build fast and highly accurate machine learning models within their existing Java application.
Alibaba, a multinational e-commerce company, uses the Native Image technology of GraalVM to statically compile microservices applications into ELF executable files which results in faster native code startup times for Java applications. Their engineering team is deploying a number of SOFABoot applications compiled as native images, and made several contributions to the project.
The Oracle Cloud (OCI) Monitoring service, a health monitoring tool, now runs on GraalVM Enterprise in production. By using GraalVM, the Monitoring service reduced its garbage collection times by 25%, application pause times by 17%, and saw a 10% increase in throughput. The benefits of these improvements are being felt across the entire Oracle Cloud platform.
Odnoklassniki is a social network service for classmates and friends. To modernize their front-end Java application to provide a good user experience with poor Internet connections they added server-side React.js rendering. After comparing implementation options they choose GraalVM for its high performance multi-language support and ease of migration.
Goldman Sachs, a multinational investment bank, is reimplementing their in-house Slang programming language used for critical pricing and risk applications on GraalVM. They used the Truffle language implementation framework to modernize Slang while not breaking its complex dynamic type system, old C++ code base, and more than 150 million lines of often mission-critical code.
Integrating GPU-accelerated libraries into existing software stacks can be challenging, especially for applications written in high-level scripting languages. Built on the Truffle language implementation framework, grCUDA lets developers efficiently share data between GPUs and GraalVM languages (Python, R, Ruby, JavaScript) and launch GPU kernels.
GraalVM’s multi-language support allows Dutch Police data scientists and application developers to collaborate by running R from Scala in a microservice architecture. With GraalVM, data scientists can focus on building statistical analysis functions in R that the service development teams can expose via Spring Boot written in Scala and Java.
GraalVM has been downloaded by developers more than 1,200,000 times to potentially improve their code and make their life easier. Why don’t you?