Public & Private Cloud Data Centers alike rely upon the high availability of applications to meet the SLAs demanded by internal and external customers. Resiliency & Reliability are important underpinnings for Data Centers to deliver high performance and high availability services. Achieving robustness in a network requires a comprehensive approach to end-to-end data integrity and link availability. Data corruption in the network can lead to rapid application performance degradation as corrupted packets are dropped and hosts are forced to retransmit or, worse yet, application bandwidth gets throttled significantly as congestion control protocols kick-in in response to packet drops and TCP session timeouts.
Innovium’s TERALYNX 7 offers the industry’s most robust network switch silicon solution with comprehensive data protection engineered in the foundation of the chip’s ground-up architecture including high levels of memory redundancy, pervasive ECC on all data structures, uncompromised protection of the entire data path and forwarding databases and, most importantly, ACL memories which are vital for security policy enforcement hardware. An additional level of protection for multi-bit errors is provided by our SDK software to ensure fast and seamless data recovery and no interruption to traffic forwarding. In order to achieve high-speed link integrity, Innovium’s TERALYNX 7 has completed robust interoperability testing across a broad 50G PAM4 ecosystem including DAC, ACC, Optics and silicon solutions to ensure that customers get the highest level of reliability. Innovium’s TERALYNX 7 is in volume production after having successfully completed exhaustive qualification testing including HTOL, PVT, and thermal chamber testing and has demonstrated robust, error-free operation.
Extensive & Comprehensive Interop Validation
While Ethernet does have a robust built-in error detection mechanism, namely the CRC check, Electromagnetic Radiation, Radio Frequency Interference, Silicon subcomponent induced alpha/beta radiation or high energy particles from cosmic radiation in the form of neutrons are some of the persistent and intractable causes that can result in data corruption, specifically while the packets are being processed through the switching silicon. Enterprise class servers have long recognized the need for robust data protection against these causes and have ubiquitously deployed ECC protection on server memory to achieve reliability. For data centers located at higher altitudes, data protection become even more critical. Studies have shown that the rate of neutron flux is 3.5 times higher at 1.5 km  altitude. Innovium’s TERALYNX 7 silicon has been rigorously tested by our system partners with neutron radiation to ensure its resiliency to data corruption.
Data Center Operators are racing to achieve quick time to revenue as well as outpace competition. Velocity of innovation enablement is paramount. High service availability makes it imperative that as innovative capabilities are rolled out, service interruptions are kept to a minimum. As new software updates are made to roll out new features or address security patches, fast recovery time is essential to ensure that control plane protocol states do not time out and that the data plane continue to forward traffic reliably and without interruption. In order to achieve this, network elements must support software upgrades without interrupting network traffic. Innovium’s chip and software’s ground-up architecture has been designed to deliver this level of resiliency. Judicious separation of data plane and control plane functionality, novel hardware acceleration in TERALYNX 7 to support fast bulk table updates and an SDK software architecture designed to quickly resume with proper state after upgrades are the key hallmarks of Innovium’s unique approach. Innovium SDK is highly robust and proven with successful NOS porting and QA efforts undertaken by multiple partners/customers.
Customers can deploy TERALYNX 7 switch silicon with high confidence backed by its proven highly reliable and resilient HW & SW. Cisco has successfully validated its robustness and FCSed multiple switches, including Nexus 3432D-S and Nexus 3408-S. Many key cloud customers have completed POC’s and are now ramping deployments. Innovium has delivered a hardened SDK and software with successful completion of PTF tests in SONiC including warm reboot, meeting SONiC’s stringent service availability requirements. Innovium leads the industry in providing the most reliable and resilient switch silicon solutions.
If you are building systems for large data centers, or operate large data centers, we’d love to show how TERALYNX 7 can help you boost resiliency and reliability of your network. Please contact us at [email protected]