Azul Systems | Unbound Newsletter October 2006 Issue

Silicon.com
Agenda Setters 2006

Silicon.com names Gil Tene, Vice President of Technology and CTO, Co-Founder of Azul Systems, as one of the top 50 most influential individuals in technology.

Read more about Gil Tene »

Horse Vapour, Anyone?

Blog by Peter Holditch,
Sr. Systems Engineer of Azul Systems

Perhaps bizarrely, this was all called to mind by a feature of the Azul Virtual Machine that I see an increasing number of customers fall head over heels in love with at first sight, the Real Time Performance Monitor. This browser-based tool allows you to peer right inside the workings of the Virtual machine engine... Application not scaling? Take a look at a thread-dump in the browser. (OK, so far, you could have got that with kill -3) lots of threads waiting for the same lock? See at a glance how long each individual thread has been kept waiting (try getting that from kill -3) Look at the stack traces, and see if there are code paths followed whilst holding locks that could be shortened... Doing I/O whilst holding a lock, go to the back of the class! Now, can't increase the throughput by adding threads? take a look at all the monitors page, see how long has been spent waiting on each lock in the VM, take the most waited for one and try to optimise that (maybe make a conventional pessimistic lock into a read-write lock, if OTC has not done it for you automatically)and so it goes on... All with minimal performance impact; all available to turn on in production or development as required. Visibility like you never had before! And when you're done with the locking behaviour, why not take a glance at the heap? Plenty to chew over for everyone!

December 5-7, 2006 | San Francisco, CA
Collaborative Communications
Summit »

Network Attached Processing Enabled by Azul Compute Appliances

What is Network Attached Processing?

Optimized for service-oriented applications, network attached processing delivers centralized compute and memory capacity to applications built on the Java™ platform in the same way network attached storage provides centralized storage capacity. Applications launch from traditional servers and seamlessly gain access to massive amounts of processors and GBs of heap space.

Network attached processing delivers these key benefits:

300% productivity improvement without any change to in-place resources
Better service levels with consistent, faster application response times
300% increase in asset utilization with lower power and cooling requirements
50% lower infrastructure cost and complexity

Enabled by Azul Compute Appliances

Network attached processing is enabled by Azul Compute Appliances. These enterprise devices are installed on a Gigabit Ethernet subnet in your application tier. Java workloads are offloaded from traditional servers via the Java-compatible Azul Virtual Machine, included with the appliances. No code changes are required for the applications to offload to the Azul platform. Once offloaded, the application gains access to larger memory heaps and as many processor cores as the application needs. This access is controlled via the Azul Compute Pool Manager, which uses customer-set policies to govern resource allocation.

Each Azul Compute Appliance is a shared resource into which multiple applications can simultaneously tap.

Each application is fully isolated from each other according to Java security rules. There are no partitions, and resources are allocated on demand to applications based on real-time workload and dynamically reallocated as needed, every 10 milliseconds.

Applications from heterogeneous hosts running different versions of Java (1.4 or 1.5) can all tap into Compute Appliances at the same and support Linux, Solaris, HP-UX and AIX operating systems.

Download the Latest Azul Case Studies