Network Attached Processing

The rapid adoption of Java™ and J2EE™ platforms and other virtual machine-based applications has caused a proliferation of application host servers in the data center, presenting IT organizations with the increasingly complex challenge of cost effectively provisioning and managing hundreds of servers without compromising system availability, utilization, and service levels.

To meet this challenge, Azul Systems introduces network attached processing, an innovative way to deliver massive amounts of compute power to enterprise Java applications, enabling organizations to achieve the same consolidation benefits in the application tier that they are already enjoying in the web and data tiers.

Network attached processing allows virtual machine based applications that are initiated on a general-purpose application host server to tap the processing power of a specialized compute appliance within a compute pool. Instead of running on the host server, the virtual machine and the application are shifted to the compute appliance for execution while all interfaces to clients, databases, and host operating system services remain on the host server. This innovative capability is enabled by Azul Virtual Machine technology.

Applications running on the compute appliance functions exactly as they would on a host-based JVM. The only difference is in the amount of available resources and the improved processing efficiency that is achieved through network attached processing. With capacities of up to 864 processor cores and 768 GB of memory in a single, coherent shared memory system, compute appliances enable creation of an application processing infrastructure that can handle fifty or more applications and thousands of concurrent process threads.

Network attached processing is a new and innovative approach to data-center management, delivering massive compute power to business applications while reducing the management complexity of today’s server farm. By providing traditional host servers with a shared compute pool of mountable ultra-high capacity Azul Compute Appliances, network attached processing provides virtually unlimited processor and memory resources for the Java-based applications.

Vega 3 System Architecture

Azul compute appliances do not run traditional general purpose operating systems. The integrated system software provides the appliance the management  and execution environments for virtual machine engines. The Azul integrated system software was designed with three key objectives:

• Enable throughput scalability to hundreds of concurrent threads

• Provide real-time response to rapid changes in resource demand to ensure consistent response times

• Enable fine-grained allocation and control of application resources

The breakthrough technology components that make up the Azul Vega 3 Solution are:

Azul Vega Processor - The first 54-core chip designed and optimized for Java workloads. The Vega chip includes features not found in conventional processors, enabling a variety of optimizations that would otherwise be not possible. With support for features such as read and write barriers that help optimize garbage collection and object relocation, speculative locking to enable safe concurrent execution of code that would otherwise be serialized, and an instruction set designed for the needs of virtual machines, the Vega processor is designed to provide consistently high throughput to Java applications. In addition, Azul Vega Processors have enterprise grade Reliability, Availability and Serviceability (RAS) features designed into them, such as, predictive failure analaysis for de-configuration and system memory chip-kill technology

Azul Virtual Machine (AVM)  - was designed to scale applications, while efficiently leveraging the capacity of the Azul Vega Appliances. The AVM provides 64-bit memory address space, even for applications originating from 32-bit hosts. It allows individual application instances to scale to 670 GB of memory and hundreds of processor cores, without changing the application code. This breakthrough scalability is achieved with unique capabilities such as hardware-assisted Pauseless Garbage Collection, which eliminates the impact of garbage collection-related application pauses, and Optimistic Thread Concurrency which minimizes the impact of scalability bottlenecks caused by synchronization and lock contentions.

Compute Pool Manager (CPM) - is a fully-integrated software application that allows administrators to manage a set of individual Vega 3 appliances as a single pool of resources. Azul CPM is a flexible, policybased management application that provides control of compute pool resources by dynamically allocating and redistributing compute resources within a pool to the various Java applications. It also provides a consolidated view of all the resources and applications in the pool to manage them from a single console. In addition, it lets users snapshot and track historical appliance capacity utilization and application resource usage.

REALTime Performance Monitor (RTPM) - is a diagnostics and tuning tool instrumented into the AVM and offers a uniquely detailed view of all performance aspects of an application, including thread, memory, and IO profiles. RTPM is always-on and has zero performance overhead. Hence, it is the ideal tool to use in production deployments to identify performance issues as they happen and take rapid corrective measures. Since it is integrated into the AVM, there is no set up required and the information is always available. 

Breakthrough Scalability and Performance

Azul Vega 3 Series appliances deliver breakthrough scalability as demonstrated with IBM Trade6, Enterprise Service Bus, XMLMark and SPECjbb2005 performance benchmarks. On these industry-standard benchmarks which are representative of typical Java workloads, the Azul model 7380 has achieved record-setting performance results.

IBM Trade6 Benchmark

IBM Trade6 (formally IBM WebSphere® Performance Benchmark Sample) is a stock trading application designed and developed to cover the significantly expanding programming model and performance technologies associated with real-world workloads, enabling performance research and verification test of the Java 2 Platform, Enterprise Edition (J2EE) 1.4 implementation including key performance components and features.²

Enterprise Service Bus Benchmark

Azul Systems tested Enterprise Service Bus scalability as measured by transactions per second, which is the critical measure for service-oriented architecture (SOA) deployments. Scalability is paramount for this pivotal SOA component to handle the fast growing and unpredictable workloads of the numerous SOA services it supports. The test measures the highest sustainable number of transactions per second (TPS) achievable on a given platform as a function of the number of threads executing in parallel (Concurrency).

XMLMark Benchmark

XMLMark is an XML parsing benchmark originally created by Sun Microsystems. The test simulates a multi-threaded server program that processes multiple XML documents in parallel. This is very similar to an application server that deploys web services and concurrently processes a number of XML documents that arrive in as client requests.³

SPECjbb2005 Benchmark 

SPECjbb2005 Benchmark is the industry standard Java server benchmark for evaluating the performance of server-side Java applications by emulating a three-tier client/server system (with emphasis on the middle tier). The benchmark exercises the various components of the JVM (Java Virtual Machine), including the JIT (Just-In-Time) compiler, garbage collector, thread management. It also measures the performance of CPUs, caches, memory hierarchy and the scalability of shared memory processors (SMPs). SPECjbb2005 provides a new enhanced workload, implemented in a more object-oriented manner to reflect how real-world applications are designed and introduces new features such as XML processing and BigDecimal computations to make the benchmark a more realistic reflection of today's applications.

The Azul Vega 3 system, featuring up to 864 processor cores and 768GB of memory in a single system, delivers the industry's highest single instance score¹ 1,507,725 SPECjbb2005 bops (business operations per second) and 1,507,725 SPECjbb2005 bops/JVM, demonstrating the unique ability to address the increasingly demanding needs of enterprise business critical Java applications. This new benchmark result is over 31% higher than the previous record.

SPEC® and SPECjbb2005® are registered trademarks of the Standard Performance Evaluation Corporation (SPEC). Competitive claims reflect results published on www.spec.org as of June 15, 2008. For details of the score, visit SPECjbb2005: http://www.spec.org/osg/jbb2005/results/res2008q2/jbb2005-20080602-00501.html.

Footnote:

¹ Competitive claims reflect results published on www.spec.org as of June 15, 2008.
² Trade6 Throughput comparisons are based on results using IBM Trade6 application
³ Based on average performance across SAX1, SAX2 and SAX3