Application Performance and Scaling Capabilities

Azul Systems unique application performance and scaling capabilities enable Java™ applications to achieve record-level workload throughput while delivering consistent response times. Existing applications perform well beyond their current limitations, and new software possibilities open up, pushing the scale even farther.

Massive Application Scalability with Predictable Response Time

The AVM overcomes application scalability and memory management limitations commonly found in Java environments. Azul innovations in the AVM and compute appliance hardware enable even the most demanding multi- threaded, high-allocation-rate applications to scale and maintain consistent response times.

Optimistic Thread Concurrency
Optimistic Thread Concurrency is an innovative lock management scheme that improves throughput by letting multiple threads execute synchronized code concurrently, detecting and rolling back any conflicts. Because actual conflicts are typically very rare, the end result is improved parallel thread execution even of serially designed code, with the overall effect of a significant improvement in application throughput. Existing programs and commonly used container libraries typically benefit from this technology and can scale to large numbers of processor cores unmodified. Additionally, developers can now program simple coarse-grained locks instead of spending time developing and debugging fine-grained locking mechanisms.

Pauseless Garbage Collection
Current Java platforms invariably bring applications to a complete halt to clean up memory space, driving transaction response times above expected levels and compromising quality of service. Azul Compute Appliances eliminate virtually all application pauses associated with garbage collection by using a unique hardware-assisted algorithm. Pauseless Garbage Collection is concurrent to the application, always able to distribute free memory to threads; and is parallel, scaling to additional available cores to keep up with soaring allocation rates. The elimination of unruly pauses means applications can deliver consistent, predictably fast response times and drive more revenue through improved SLAs.

600+ GB of Pauseless Heap Memory
Exclusively on Azul, Java application heap memory can safely expand to 670 GB with no garbage collection pauses. This is the only viable capability on the market enabling consistent throughput ramp up with a single instance of the application allowing for fewer cluster nodes, wider data sets for more sophisticated calculations, and larger data caches.

DirectPathVM™
In highly modular application architectures, such as those typically found in SOA (service oriented architecture) and EDA (event driven architecture) deployments, the end-user’s Quality of Service can be significantly impacted by I/O latency between multiple application services. By consolidating application services on a pool of Azul Compute Appliances, applications avoid the bottlenecks associated with network I/O. Azul DirectPath provides a high bandwidth interconnect that eliminates I/O bottlenecks between application services. The resulting decrease in transaction response time and improved transaction throughput yields much higher quality of service for the end-user.

REALTime Performance Monitor
REALTime Performance Monitor (RTPM) is a diagnostics and tuning tool instrumented into AVM. RTPM offers a uniquely detailed view of the real-time activity of both the application and the underlying VM. RTPM is the perfect tool for rapidly diagnosing scalability, response time and throughput problems in record time.

• Threads View. Shows how long threads have been waiting and the culprit threads. Snapshots all thread states and stack traces. Clarifies what the application is doing at every moment, and lets users trace inefficient synchronizations

• CPU Usage Profiler. Identifies execution hotspots as candidates for optimization, shows activity level of application or JVM components over time, and helps assess parameter tuning.

• Hot-locks. Identifies the problematic locks costing the program the most in time and effort, allowing programmers to reduce contention so application threads can zip forward.

• System call profile. Displays frequent system calls that demand the longest time to complete, enabling better mitigation of impact on application progress.

• No-setup. RTPM requires no setup, preliminary code analysis or filtering. It instruments the entire application, making all information ready for user consumption and examination.

• Lightweight and always-on. RTPM information is collected nonstop with minimal execution impact, and can be confidently used in real-time on production applications without stability or performance concerns. RTPM offers IT organizations the power to analyze application behavior under real-world loads and gives developers insight to design issues without compromising service level agreements.