The Silicon Review
We offer the broadest range of software compatible data center switches ranging from 3.2Tbps to 12.8Tbps with leading analytics, programmability and power efficiency: Rajiv Khemani, CEO & Co-founder Innovium Inc.
Enterprise customers are rapidly moving to hybrid cloud deployments, using both private and public could data centers. IDC forecasts that by 2018 more than 60% of enterprise IT organizations will have committed to multi-cloud architectures, driving up the rate and pace of change in IT organizations. Mark Hurd, CEO of Oracle, has predicted that cloud data centers will dominate a significant majority of the $1 trillion enterprise IT economy, and that at this time we are in the first innings of the transition with the cloud representing only $50 billion.
As a result, large data centers are evolving their infrastructures to meet the growing needs for the next generation of IT/cloud services. They need the right compute, storage and network infrastructure in place to meet their growing performance and scale needs, simplify operations using analytics and automation, future proof the infrastructure for newer types of IT/cloud services, and make them more power efficient.
In view of the above mentioned trend, we are excited to present Innovium Inc., a leader in enabling autonomous and programmable large-scale data center networks.
The firm designs and develops Ethernet switch silicon solutions for data centers customers. It offers game-changing networking infrastructure solutions to meet the critical needs of these customers.
The company was started in January 2015 and is based in San Jose, California.
Interview Excerpt: Rajiv Khemani, CEO & Co-founder
Why was the company set up?
Network traffic in large data centers is continuing to see explosive growth, propelled by huge demand from cloud computing, artificial intelligence/machine learning, and data analytics to high resolution streaming video, micro-service based applications, and distributed storage. To address these demands on the network, large data center customers need to deploy optimal network architectures in order to provide the best application performance while lowering operational expenses through lower energy consumption, simplicity, and automation. As the scale of data-centers increase, the need for software-driven programmability, advanced telemetry, and guaranteed high bandwidth fault tolerant network becomes an even more important requirement.
Recognizing this, we founded Innovium with the vision to deliver the world’s highest performance, most innovative Ethernet switch silicon optimized for next-generation data centers. Innovium started with a ground-up design which is not encumbered by legacy architectures.
Tell us about your first product that was launched.
Innovium TERALYNX is fuelled by customer-focused innovation. It delivers the world’s fastest and most scalable Ethernet switch silicon family with industry-leading analytics, programmability, and power efficiency. The TERALYNX 7 family offers up to 12.8Tbps single-chip performance; 2x performance per watt and 3x buffer size versus current alternatives; breakthrough telemetry and analytics; and support for new 200G and 400G Ethernet standards. The modern and scalable TERALYNX switch silicon architecture offers highly innovative data center focused capabilities in a modular and area-optimized fashion to scale from 2Tbps to 51.2Tbps and beyond to address the bandwidth needs over the next decade.
There are three key areas of innovation INNOFLEX, TERASCALE, and FLASHLIGHT.
In addition, TERALYNX features industry-leading, proven SerDes that supports 10G and 25G NRZ as well as 50G PAM4 I/O standards. This provides customers a variety of connectivity choices ranging from the widely deployed 10/25/40/50/100G Ethernet to the upcoming 200/400G Ethernet.
What can you tell us about your product offerings?
Innovium TERALYNX™ Ethernet Switch Silicon family delivers the world’s highest performance switch silicon with large buffers, unmatched analytics through fine-grain telemetry, line-rate programmability, the industry’s best power efficiency in terms of performance per watt and very low latency. The TERALYNX product family offers the broadest range of data center switches ranging from 3.2Tbps to 12.8Tbps, enabling compelling and unique systems such as 128 ports of 100GbE, 64 ports of 200GbE and 32 ports of 400GbE.
What do you feel are the reasons behind:
Your product popularity – Innovium TERALYNX has been designed from ground up to optimally address the needs of large data center customers. From day one, our team focused its attention on data center customer pain points and trend lines and oriented its innovation around addressing those problems. Our customer-focused innovation has resulted in over 40 patents.
Your consistence growth as an organization – We have an unmatched, world-class engineering and executive team with prior experience at leading technology companies that include Arista, Broadcom, Cavium, Cisco, Dell, Ericsson, Intel, and Juniper. The team has proven experience in defining and delivering a diverse set of networking products that have been pervasively deployed by the most demanding cloud and enterprise customers and have generated billions of dollars in revenues.
Where do you see your company a couple of years from now?
We expect our products to be deployed across multiple data center customers worldwide, providing best-in-class performance, programmability, buffers, latency and advanced telemetry to enable autonomous scalable data center networks with the very compelling total cost of ownership (TCO).
Rajiv Khemani, CEO & Co-founder: Mr. Rajiv Khemani has over 20 years of business management and leadership experience in start-up and public companies. Previously Mr. Khemani was COO of Cavium, a leading cloud and mobile infrastructure solutions company. At Cavium, he held executive roles over a 10-year period leading marketing, sales, engineering, and operations, helping the company achieve a successful IPO and a multi-billion dollar valuation. Mr. Khemani also spent 5 years at Intel, where he was GM of Intel’s network processor business. He joined Intel as a result of the acquisition of Net boost, a network processor start-up. Earlier in his career, he worked at Net app and Sun Microsystems. Mr. Khemani holds an MBA from Stanford University, an MS CS from NYU and a BS in Computer Science & Engineering from IIT Delhi.
Innovium was co-founded by Rajiv Khemani, Puneet Agarwal and Mohammad Issa. Prior to Innovium, Puneet and Mohammad worked at Broadcom for approximately 10 years leading architecture and engineering organizations in the switch group. Innovium team members have previously defined, designed and delivered multiple generations of highly successful products that are deployed by the most demanding cloud and enterprise customers and have generated billions of dollars in revenues.