Tervelaâs test results are millions of messages per second higher than any previously released public vendor test data. The tests, which were performed at Advanced Micro Devices (AMD)âs Sunnyvale, California headquarters, represent the first time that vendor claims about message processing speeds and latency have been publicly demonstrated and validated using a large-scale environment consisting of over 100 high-performance servers to emulate realistic market loads.
The test conditions were constructed using parameters that best matched the operations of most large financial services firms including Gigabit Ethernet, 150 byte messages and UDP point-to-point transport.
âWhen performing benchmark tests on latency and throughput, vendors will often create custom environments, use unrealistically small messages or eliminate data outliers when calculating resultsâ, said Barry Thompson, Founder and CTO, Tervela. âOur industry needs to find a way to standardize these test environments so end users can make apples to apples comparisons when making technology purchasing decisions. The last thing they want is to be surprised by devastating performance issues when messaging solutions are deployed in live environments.â
âThe Tervela benchmarks in the AMD lab represent a leap forward in how we assess and measure systems that will be deployed in the worldâs most demanding IT environments,â said John Fritz, director of Strategic Software Alliances, AMD. "The importance of real-world data traffic results is significant, especially for those in the financial services industry. Tervelaâs outcome is a great example of the type of landmark improvements in testing accuracy and relevance that can be achieved leveraging AMD technology.â
The Tervela Message Network is a hybrid of networking and middleware technologies built around a message switch. Features include:
â¢ Ultra-low latency â even during peak volatility periods when software and legacy solutions experience huge spikes in latency
â¢ Predictability â of operational performance that allows data-intensive applications to ingress, process and disseminate critical information deterministically
â¢ High capacity â intelligently filter and route data in a cloud configuration that can scale to tens of millions of events per second without data loss
â¢ High fan-out â to hundreds of subscribers without the complexity, reliability or manageability issues of legacy approaches
â¢ Efficiency â to reduce data center footprint, streamline management controls and optimize network bandwidth for messaging