Azul Systems®, the award-winning leader in Java runtime scalability, today announced the results of a data distribution benchmark performed by Push Technology, an Application and Integration Gartner Cool Vendor best known for its leading high-performance, scalable and real-time data distribution solution, Diffusion™. The results show convincingly that Zing's performance is unmatched for applications and service level agreements (SLAs) that require high throughput and consistently low latency. Compared to a legacy JVM, Zing was able to handle 3x more connections while simultaneously reducing peak latencies by over 67x.
Push Technology, using their open source benchmark suite allowing the performance results from the study to be easily reproduced, undertook the benchmark to compare the performance and latency differences of Java Virtual Machines (JVMs). The benchmark was run on Diffusion software on two multicore
Diffusion itself is a real-time stream oriented distribution technology that intelligently batches, fragments and conflates messages to optimize distribution whilst guaranteeing the currency, and timeliness of delivery. Push Technology's Diffusion servers are typically deployed at or behind the edge of private networks where services are exposed to public networks and real-time data feeds, with multiple instances running behind a load balancer.
Comparative latency graphs show that throughout the Diffusion tests Zing's response time was consistently and favorably flat -- exhibiting stable, consistent and delay-free operation with a measured worst case peak latency of 1.25 milliseconds - while simultaneously supporting the full load of 15,000 simultaneous connections. By contrast, running with the legacy JVM showed peak pauses as high as 70-80 msec, with significant stalls ranging into the 10s of milliseconds occurring as early as the 99th percentile. In contrast to Zing's ability to maintain a flat latency curve at all loads, the alternative JVM was unable to avoid these frequent and high magnitude stalls and peaks even when carrying only 5,000 (3X fewer) live connections.
Solarflare VP of Solutions Marketing
A detailed summary of these benchmark results is available on the Push Technology Web site at technology.com/azul">http://more.pushtechnology.com/azul. The Open Source Push Technology Diffusion™ benchmark is available on github.
Note for editors: Java is deployed throughout many industries in latency-sensitive, revenue-critical applications ranging from real-time wireless billing to algorithmic trading and social gaming -- and both established and emerging businesses continue to place new demands on their applications for high levels of overall performance as well as consistent, jitter-free operation. In many cases Java meets these demands, but intermittent Java-based performance stalls, glitches, and pauses can pose a risk to real-time business operations, impacting revenue, customer loyalty and developer productivity. These results highlight how changing from legacy Java runtimes to Zing can have a direct impact on application performance and peak latency.
Most Popular Stories
- Slow Week Ahead of December FOMC Meeting
- Hispanics Seek to Grow School Board Members
- GM Bailout Saved 1.2 Million U.S. Jobs, Report Says
- Bitcoin Used to Buy Tesla Car
- U.S. Companies Eager for Iranian Business
- 'Knockout Game': Myth or Menace?
- Banks Fret as Volcker Vote Approaches
- Questions Remain in Jenni Rivera's Death
- Yellen Set to Become One of World's Most Powerful Women
- Paul Walker Fans Pay Respects