Breaking News - Operating Systems & Middleware:
Azul Introduces First Network Attached Processing Solution
Azul Systems Inc introduced the industry's first network attached processing
solution designed to provide unbound compute resources for Java and J2EE
platform-based applications, a growing need according to industry analyst firm
Gartner Group. Gartner estimates that by the end of 2008, more than 80 percent
of all new eBusiness application development will be based on virtual machine
environments, including the Java and J2EE platforms and the Microsoft .NET
Framework. Driven by this change, Azul Systems developed its network attached
processing solution to reduce the cost and complexity of delivering and
managing compute resources in the most demanding data center environments.
"To date, server-side Java deployments have generally been constrained by CPU
costs, the lack of large-scale parallel execution within a virtual machine,
and the overhead of provisioning new Java applications to respond to peak
demand," said Scott Dietzen, former CTO of BEA Systems. "Azul Systems compute
pools have the potential of improving scalability and performance for today's
application server platforms from Apache Tomcat to BEA WebLogic and IBM
WebSphere as well as for Java enterprise applications."
The Azul Systems network attached processing solution will be delivered via
three unique technologies:
- Mountable pools of compute for Java platform-based applications enabling
any application to draw computing capacity from the same massive enterprise
resource.
- Flexible, policy-based management capabilities built into the compute
pools, providing comprehensive command and control of compute resources.
- Transparent virtual machine proxy technology redirects application
workloads to the compute pool without requiring any changes to applications or
existing infrastructure.
Azul Systems compute pools are comprised of ultra-high capacity compute
appliances. Based on innovative multi-core silicon technology, each compute
appliance contains up to 384 coherent processor cores and 256GB of memory in a
fully symmetrical multiprocessing (SMP) system. Azul compute appliances are
specially optimized for object-oriented languages and virtual machine based
applications. A standard datacenter rack can house several appliances and more
than 1,200 processor cores. Azul compute pools can be tapped by any Java or
J2EE platform-based application in the datacenter without any application
modifications, long-term binary or instruction set lock in. Elegantly
packaged, this targeted design provides a small footprint, high rack density,
low power consumption and simple administration.
With the introduction of network attached processing, which has roots in
similar changes that occurred in storage and networking, the Azul Systems
compute appliances will transform enterprise compute resources into virtually
unbound pools of compute for J2EE platform-based applications built on top of
today's leading application server platforms. This new approach does not
require any application-level modifications, binary compatibility requirements
or operating system dependencies. Network attached processing provides
massive, shared compute capacity that can dynamically respond to changing
workloads, eliminating capacity planning at the individual application level.
By simply plugging Azul compute appliance into the network, customers will see
dramatic improvements in performance and reduce overall IT costs and
management complexity.
"Customers are transitioning from legacy development environments like C, C++
and COBOL to application virtual machine environments like Java, J2EE and
.NET. These modern development environments expose a highly parallel,
multi-threaded workload to underlying hardware. This opens up an opportunity
to develop a new class of systems that solve real application scaling issues
in an easy to manage fashion," said Stephen DeWitt, president and CEO of Azul
Systems. "Our vision is not to reinvent the past, but to look toward the
future and fundamentally change the manner in which the function of compute is
delivered and its associated economics."
|