Wednesday, July 19, 2023

Ultra Ethernet Consortium targets networking for AI and HPC

A new Ultra Ethernet Consortium (UEC) has been established with the goal of bringing together leading companies for industry-wide cooperation to build a complete Ethernet-based communication stack architecture for high-performance networking. UEC is 

The aim is to capitalize on Ethernet's ubiquity and flexibility for handling a wide variety of workloads in Artificial Intelligence (AI) and High-Performance Computing (HPC).

Founding members of Ultra Ethernet Consortium, which is a Joint Development Foundation project hosted by The Linux Foundation , include AMD, Arista, Broadcom, Cisco, Eviden (an Atos Business), HPE, Intel, Meta and Microsoft.

The consortium will follow a systematic approach with modular, compatible, interoperable layers with tight integration to provide a holistic improvement for demanding workloads. The founding companies are seeding the consortium with highly valuable contributions in four working groups: Physical Layer, Link Layer, Transport Layer and Software Layer.

The technical goals for the consortium are to develop specifications, APIs, and source code to define:

  1. Protocols, electrical and optical signaling characteristics, application program interfaces and/or data structures for Ethernet communications.
  2. Link-level and end-to-end network transport protocols to extend or replace existing link and transport protocols.
  3. Link-level and end-to-end congestion, telemetry and signaling mechanisms; each of the foregoing suitable for artificial intelligence, machine learning and high-performance computing environments.
  4. Software, storage, management and security constructs to facilitate a variety of workloads and operating environments.

"This isn't about overhauling Ethernet," said Dr. J Metz, Chair of the Ultra Ethernet Consortium. "It's about tuning Ethernet to improve efficiency for workloads with specific performance requirements. We're looking at every layer - from the physical all the way through the software layers - to find the best way to improve efficiency and performance at scale."  

“The next era of computing will be characterized by breakthrough advancements in AI and AI-optimized infrastructure, and Microsoft is committed to empowering organizations to push the bounds of what is possible with the power of Azure. Joining forces to develop a common set of standards to enhance Ethernet for hyperscale AI and high-performance computing workloads will help enable continued innovation now and in the future,” said Steve Scott, Corporate Vice President of Azure Hardware Architecture at Microsoft.

Wondering whether to use Infiniband or Ethernet for building large-scale clustered GPU environments?

Juniper powers BT Network Cloud

Juniper is providing the high-performance, high-availability fabric for BT Network Cloud, which is the infrastructure that underpins BT Group's U.K. consumer and business network services.

BT Network Cloud supports diverse network applications across broadband, fixed, mobile and other mission-critical services on a single platform, using Juniper Networks’ high-performance cloud-native solutions and professional services.

BT’s cloud-native platform is deployed at multiple locations across the U.K. to form a distributed architecture designed to scale seamlessly alongside demand and to provide services closer to the end user than with traditional, more centralized deployments. The fabric is now supporting multiple key network-centric applications, such as mobile voice and data, broadband authentication and IP voice calls for tens of millions of customers.


  • Software-defined networking (SDN), switching and firewall solutions for its network cloud. The service provider uses Juniper’s SDN platform to automate the creation and management of its OpenStack-based telco cloud. The platform provides hybrid SDN orchestration and centralized control of virtual switching, routing and security. BT is running cloud-native containerized applications and optimizing costs using OpenStack.
  • Juniper QFX Series Switches provide the high-performance, high-availability fabric needed to support the BT Network Cloud which has rapidly become the largest container-based cloud infrastructure in the world, and is still growing.
  • Juniper collaborated with BT to deliver high-performance, high-availability fabric for BT Network Cloud. BT’s data centers contain over 1,000 Juniper QFX Series Switches and 2,000 Juniper vRouters. The switches connect all mission-critical application servers to the core network. The SDN platform is deployed on the servers to provide virtualized routing capability to connect workloads across BT’s network securely.
  • Juniper vSRX Virtual Firewalls protect and secure the data center traffic. vSRX provides a network address translation capability which was not possible for one of BT’s application providers to deliver in software.

Juniper noted that BT Group leveraged its professional services to complete the project on time and ensure that the challenging service level, availability and performance targets of BT’s applications are met.

“Telecoms services have become central to business and consumer lives and as the U.K.’s national network provider, BT Group has a particular responsibility to lead from the front to deliver the cloud-native, sustainable digital services of the future. Our work with partners such as Juniper is helping to ensure that our critical infrastructure, network and services are agile, secure and resilient for the long term. Our millions of customers can remain confident in the experience-led integrity of the everyday and exceptional digital services on which they rely.,” stated Howard Watson, Chief Security and Networks Officer, BT Group.

“We are delighted that Juniper’s cloud-native networking solutions and comprehensive professional services were selected as the cornerstone of the BT Network Cloud. BT can respond to the demands of the next chapter of digital transformation at scale whilst leveraging the power of the network to enable new digital services for BT’s customers, fully supported by a cloud-native architecture enabling security, automation, scale and agility,”states Raj Yavatkar, Chief Technology Officer, Juniper Networks.

Zayo adds long-haul fiber routes, 400G services, data center PoPs

Zayo announced a series of expansions to its fiber infrastructure, including new long-haul dark fiber routes, increased fiber capacity for existing routes, additional 400 Gbps (400G)-enabled routes and new IP points of presence (PoPs).

New long-haul dark fiber routes include:

  • St. Louis to Indianapolis: Expands connectivity in these growing markets, bypassing the heavily trafficked Chicago metro area with four In-Line Amplification (ILA) sites in key markets along the route.
  • Chicago to Omaha Overbuild: Provides the most direct route between Denver and Chicago, including a new high fiber count cable that connects 11 ILA sites in strategic markets, facilitating seamless network connections for customers. As a heavily used route, this overbuild will allow for more capacity.

New 400G:

Four new 400G-enabled routes were made operational in the second quarter of 2023 and two more slated for completion in Q3

  • Las Vegas to Phoenix
  • Kansas City to Indianapolis
  • Dallas to East St Louis
  • Albany to Boston
  • Denver to Dallas (a direct route scheduled for completion in Q3 2023)
  • Atlanta to DC (scheduled for completion in Q3 2023)

IP PoPs:

Zayo added IP PoPs in 14 neutral data centers in Q1 and nine in Q2 across North America.

Enterprise Networking Alphabet Soup: Mastering Multi-Cloud Complexity with IBM

How to efficiently manage the complexities of modern enterprises across multiple environments? 

Murali Gandluru, Vice President of Strategy (Product & GTM), Software Networking Business Unit from IBM explains: 
  • How IBM Hybrid Cloud Mesh simplifies and secures access across various environments 
  • The benefits of Hybrid Cloud Mesh for DevOps organizations 
  • How IBM achieves business outcomes across all applications and domains

Full showcase on #SASE, #SSE, #SDWAN, #ZTNA, #MCN, #NaaS at:

Want to be involved in our video series? Contact

Enterprise Networking Alphabet Soup: Building Secure Global Networking at Cato

Wondering how to simplify your network security and management?

Eyal Webber-Zvik, VP Product Marketing from Cato Networks explains:

  • How Cato Networks is the world's first true converged SASE platform, combining networking and security capabilities into one globally available cloud-native platform.
  • Cato's Points of Presence (POPs) around the world, interconnected with a private global backbone (Cato's private highway) that expedites enterprise traffic instead of messy traversal over the public internet.
  • Cato's comprehensive real-time security enforcement that includes firewall as a service (FWaaS), secure web gateway (SWG), intrusion prevention (IPS), anti-malware, zero trust network access (ZTNA), CASB, DLP and remote browser isolation (RBI), with more capabilities coming shortly.

Enterprise Networking Alphabet Soup: Cisco's Cloud Solution

Struggling to simplify networking experience across multiple domains?

Kip Compton, SVP, Strategy & Business Development from Cisco Systems explains:

  • How to unify and manage SD-WAN, data center networking, cloud networking, and wireless technologies
  • How enterprises can deliver seamless and incredible experiences to users with end-to-end solutions
  • Utilizing cross-platform navigation and single sign-on for easy management across domains

Full showcase on #SASE, #SSE, #SDWAN, #ZTNA, #MCN, #NaaS at:

Want to be involved in our video series? Contact

MACOM wins DoD contract for Gallium Nitride-on-Silicon Carbide R&D

MACOM has been awarded a contract from the United States Air Force Research Laboratory to develop advanced semiconductor process technology related to Gallium Nitride-on-Silicon Carbide (GaN-on-SiC). 

The research will focus on GaN-on-SiC semiconductor process technologies used in millimeter-wave Monolithic Microwave Integrated Circuit (MMIC) products. The contract, valued at approximately $4.0M USD, will have a multi-year period of performance.

“This award supports our strategy to establish leadership in high-frequency and high-power MMIC technologies and products,” said Stephen G. Daly, MACOM’s President and Chief Executive Officer.

In January 2021, MACOM and AFRL entered a Cooperative Research and Development Agreement (CRADA) to transfer AFRL’s production ready 0.14 micron GaN-on-SiC semiconductor process to MACOM’s Massachusetts-based U.S. Trusted Foundry.

Samsung unveils Graphics Double Data Rate 7 (GDDR7) DRAM

Samsung announced the industry’s first Graphics Double Data Rate 7 (GDDR7) DRAM with a bandwidth of 1.5-terabytes-per second, yielding 1.4X boost in performance versus GDDR6, which Samsung introduced in 2022.

Samsung’s GDDR7 achieves a boosted speed per pin of up to 32Gbps. The enhancements are made possible by the Pulse-Amplitude Modulation (PAM3) signaling method adopted for the new memory standard instead of the Non-Return-to-Zero (NRZ) from previous generations. PAM3 allows 50-percent more data to be transmitted than NRZ within the same signaling cycle. 

“Our GDDR7 DRAM will help elevate user experiences in areas that require outstanding graphic performance, such as workstations, PCs and game consoles, and is expected to expand into future

 applications such as AI, high-performance computing (HPC) and automotive vehicles,” said Yongcheol Bae, Executive Vice President of Memory Product Planning Team at Samsung Electronics. “The next-generation graphics DRAM will be brought to market in line with industry demand and we plan on continuing our leadership in the space.” 

In addition, to minimize heat generation, an epoxy molding compound (EMC) with high thermal conductivity is used for the packaging material in addition to IC architecture optimization. Samsung says these improvements dramatically reduce thermal resistance by 70-percent in comparison to GDDR6, aiding in stable product performance even in conditions with high-speed operations.

Digital Realty sells 65% stake in 2 Chicago data centers for $743M

Digital Realty has partnered with GI Partners to establish a joint venture for the sale of a 65% interest in two data center buildings and  associated equipment in the Chicago metro area.  Digital Realty will receive approximately $743 million of gross proceeds and will maintain a 35% interest in the joint venture while continuing to manage the day-to-day operations of the assets, providing a seamless customer experience.  

Digital Realty has also granted GI Partners an option to purchase an interest in the third facility on the same hyperscale data center campus.

Digital Realty originally acquired the facilities in 2017 through its merger with DuPont Fabros.  The two data centers contributed to the joint venture contain approximately 67 megawatts of IT capacity and are 90% occupied in aggregate, primarily by investment grade customers.

  • In 2017, Digital Realty acquired DuPont Fabros Technology (DFT) in an all-stock transaction valued at approximately $7.6 billion. DFT owns and operates a fleet of 12 purpose-built data centres concentrated in Northern Virginia, Chicago and Silicon Valley - three markets red hot for data centre activity. The DFT properties offer a combined total 3.5 million gross sq feet and 302 megawatts of available critical load. Digital Realty is the premier name in data centres, as it operates 156 key colocation facilities in 11 countries on four continents. The merger especially boosted Digital Realty's presence in hyperscale data centres in top U.S. markets.

SiPhox raises $27M for home blood test powered by silicon photonics

SiPhox Health, a start-up based in Burlington, MA, raised $27 million in funding for its work in lab-grade home health testing kits powered by silicon photonics.

SiPhox, which was founded by MIT scientists Diedrik Vermeulen and Michael Dubrovsky, is seeking FDA clearance for its SiPhox Home platform, which will offer a broad menu of proteins and hormone tests from a finger prick blood sample, with results in five minutes or less. In the meantime, SiPhox Health has launched a mail-in blood collection kit that measures 17 biomarkers for inflammation, metabolic, hormonal and cardiovascular health.

The funding included a $10M Seed financing and a $17M Series A round. The Series A round was led by Intel Capital with participation from Khosla Ventures, Kortex Ventures, Alumni Ventures, Metaplanet, Shorewind Capital, LongeVC, Overlap Holdings and Duke Capital Partners. The Seed funding was led by Khosla Ventures and Y Combinator with participation from Metaplanet, Massachusetts Manufacturing Innovation Initiative, Rsquared, Vituity, Paul Buchheit, Balaji Srinivasan, Bob Lee, and Longevity Tech Fund among others. 

“At SiPhox Health, we are leveraging the trillions of dollars invested in the semiconductor industry to enable lab-quality results in a consumer-ready, user-friendly device,” explained Diedrik Vermeulen, SiPhox Health co-founder and CEO, and a silicon photonics pioneer who played a crucial role in developing the first coherent silicon photonic transceiver chip, now responsible for a significant share of global internet traffic.

“SiPhox’s goal is to create category-defining health tracking products starting with the SiPhox Home, which is a 100-fold improvement over existing blood diagnostics. Eventually, our technology will enable the ultimate wearable device for measuring proteins, hormones, and small molecules continuously,” said Michael Dubrovsky, SiPhox Health co-founder and Chief Product Officer. “Every cell in the human body is a much more advanced sensor than anything on the market today, showing we are nowhere near the physical limits for performance and miniaturization in diagnostics.”