GraalVM is a Technology to Watch

Oracle

This was originally published on the Amalgam Insights site on April 17, 2017.

Oracle Labs has recently (April 17, 2018) announced the 1.0 release of GraalVM. GraalVM is an open source language virtual machine(VM), much like the Java VM or Node.js virtual machine. What makes GraalVM interesting, is that it can execute code written in a variety of languages including Java (and Java VM based languages such as Scala, Groovy, or Kotlin), R, JavaScript, along with Ruby, R, and Python. This is a departure from mainstream VM designs. It is much more common to have separate and specific VMs for languages such as PHP or Python. In some cases, a language will byte compile to a different virtual machine, for instance Clojure compiling to the run on the Java VM. Those languages are purpose built to run on a specific VM. GraalVM, on the other hand, runs code written in languages originally built for their own VM.

This approach offers advantages over the traditional approach of one language, one VM. For example, any program that is compiled for GraalVM can share libraries with other programs that is likewise compiled. Developers can write in different languages but still maintain interoperability and code reuse across them all. This also allows developers to continue to use code written in “older languages” while migrating to a new one. Similarly, it allows the continued used of majority language, such as Java, while leveraging languages that are built for specific purposes, such as R. Another advantage of GraalVM is ubiquity. One VM for multiple needs means fewer VMs to provision and update across IT servers and containers. That can be a serious time saver and makes maintaining large and complex systems a bit easier.

GraalVM is also embeddable. The ability to build plugins using GraalVM was one of the reasons that Oracle started the Graal project. By creating GraalVM compatible code that can be embedded in an Oracle Database opens the door to customers extending their Oracle databases. This is clearly better than writing code outside the database for common query-related processing.

At the moment GraalVM only supports a handful of languages – in other words, it does not have ubiquitous support for all common programming languages. The mainstays of Microsoft application programming, Microsoft CLR based languages such as C# and F#, are not available. Obviously, languages that are native compiled, such as C, C++, or Go, won’t run on GraalVM unless they can be compiled for it. This will probably limit GraalVM to the Java and open source language crowd that is dabbling in Node.js JavaScript for microservices or R and Python for analytics. Another small hang-up is that Ruby, R, and Python are listed as “experimental.” This should inhibit deployment of these languages in production environments using GraalVM. That is a temporary issue that time will hopefully solve.

GraalVM has serious potential. It can simplify a complex VM landscape and allow choice in programming languages without proliferation of VMs and libraries. At the moment, it’s mostly potential, especially given the limited range of supported programming languages. That potential, however, is the reason why GraalVM is worth keeping an eye on.

Comments are closed.