Azul Systems, the Java server appliance house, says Java, Ruby and, for that matter, .NET managed runtimes are pushed to their limits and basically crippled.

Their response times are inconsistent, their scale is limited, and they’re unable to exploit modern commodity hardware with its multi-cores and memory. They’re unstable and need continuous tuning. They suffer from daily restarts, garbage collection pauses and out-of-memory errors. Operating systems aren’t optimized for them and virtualization and the cloud aggravate the scaling issue.

So Azul, which obviously has a dog in this race, is trying to drum up industry support for a broad-based Managed Runtime Initiative that would fix the problem.

It can’t do it alone because multiple layers of the systems stack have to be changed and that needs cooperation from a bunch of projects and vendors.

Azul’s effort immediately got Java inventor James Gosling to agree that the problem exists. “Managed runtimes,” he’s quoted as saying, “have come a long way since the mid-90s. However, the rest of the systems stack has not evolved to meet the needs of these now pervasive application environments. This initiative will serve to bring new functionality to the systems stack, freeing managed runtimes to continue their growth and evolution.”

Azul claims the situation can be fixed – and that performance and scalability can be improved 100x – by enhancing the interfaces between the managed runtime, the operating systems’ kernel, the underlying hypervisors and the hardware.

Its proposed open initiative is supposed to identity, develop and deliver these enhanced interfaces across the vertical components of the whole systems stack.

It wants to start with the interfaces to Linux and is throwing in an open source reference implementation that enhances OpenJDK, the open source release of the Java compiler runtime, and the Linux 2.6.32 kernel.

This 0.9 project, the work of its open source Java team and licensed under the GPL 2, includes a fully functional Java virtual machine based on an enhanced version of OpenJDK 6 coupled with a bunch of enhanced loadable Linux kernel objects that together are supposed to result in a 100x improvement in runtime execution because of pauseless garbage collection and an increase of two orders of magnitude in object allocation resulting in better scalability and consistency.

Azul says this demo widgetry is limited to supporting a few simple Java applications and that it’s working on a 1.0 rev that it’ll release in the second half. It’s hoping others will get involved.

Azul is advocating that reference implementations be used to demonstrate the value of any jimmying rather than specific APIs.

It wants to get open source communities involved and, where appropriate, there will be upstream contributions to existing and complementary open source projects like kernel.org.

The initiative should ultimately encompass multiple projects – and if they’re willing – proprietary widgetry though the effort’s likely to be dominated by open source. Azul has scheduled its recruitment effort for the second half.

It says commercialization has yet to be determined.

See www.managedruntime.org