Monday, April 24, 2023

Cadence tapes out UCIe chaplet die-to-die on TSMC 3nm

Cadence Design Systems confirmed the tapeout of its 16G UCIe 2.5D advanced package IP on TSMC’s 3nm (N3E) process technology. 

Cadence UCIe IP provides an open standard for chiplet die-to-die communication, which is becoming more critical for artificial intelligence/machine learning (AI/ML), mobile, automotive, storage and networking applications.

Cadence said it is currently engaged with a pipeline of Tier 1 customers, and UCIe advanced package IP collateral from the N3E test chip tapeout is shipping and available. The pre-verified solution can save customers time and effort through rapid integration.

The heterogeneous integration of Cadence’s UCIe PHY and controller eases chiplet solutions with die reusability. The complete solution includes the following, which can be delivered with a complement of Cadence Verification IP (VIP) and TLM models:

  • UCIe Advanced Package PHY: Designed for a bump pitch that enables greater than 5Tbps/mm of die edge bandwidth density, the UCIe advanced package PHY offers options that allow greater throughput performance while significantly improving power efficiency. It is flexible for integration on multiple types of 2.5D advanced packages, such as silicon interposer, silicon bridge, RDL and fanout-based packaging.
  • UCIe Standard-Package PHY: Options allow customers to reduce costs while maintaining high bandwidth and power efficiency. Cadence’s circuit design allows customers to design down to the lower limits of the standard’s bump pitch range to allow maximum BW/mm while also enabling longer reach.
  • UCIe Controller: A soft IP that can be synthesized for multiple technology nodes, the UCIe controller is offered in a variety of options for different target applications and enables streaming, PCI Express® (PCIe), and CXL protocols.

“The UCIe Consortium supports companies designing chiplets for use in standard and advanced packaging. We are thrilled to extend our congratulations to Cadence on reaching the tape out milestone for the advanced package test chip which uses the die-to-die interconnect based on the UCIe 1.0 specification,” said Dr. Debendra Das Sharma, chairman at the UCIe Consortium. “Member company advancements in IP (scaling) and VIP (testing) are important components in the ecosystem. When paired with participation in UCIe work groups the industry will continue to see new chiplet based designs entering the market that are based on open industry standards that foster interoperability, compatibility, and innovation.”

www.cadence.com/go/ucie16g

Cadence speeds up SoC design with 112G SerDes IP on TSMC's N4P process

Cadence Design Systems introduced its 112G Extended Long-Reach (112G-ELR) SerDes IP on TSMC’s N4P process for hyperscale ASICs, artificial intelligence/machine learning (AI/ML) accelerators, switch fabric system-on-chips (SoCs) and 5G wireless infrastructure. 

The extended long-reach SerDes PHY supports insertion loss (IL) of 43db with BER of 10e-7—thereby providing additional performance margin beyond the standard long-reach specifications—and enables exceptional system robustness for lossy and reflective channels observed in open box platforms as well as lengthy direct attach copper (DAC) cables.

The Cadence 112G-ELR SerDes PHY IP on TSMC’s N4P process, a performance-focused enhancement of the TSMC 5nm technology platform, incorporates DSP-based SerDes architecture with maximum likelihood sequence detection (MLSD) and reflection cancellation technology. The SerDes PHY IP is compliant with IEEE and OIF Long-Reach (LR) standards while providing extra performance margin for ELR applications. The optimized power, performance and area are ideal for different user scenarios, including high port-density applications. In addition to ELR and LR channels, the IP also supports Medium Reach (MR) and Very Short Reach (VSR) applications with a flexible power-saving capability over different channels. The supported data rates range from 1G to 112G with NRZ and PAM4 signaling, enabling reliable high-speed data transfer over backplane, direct-attached cable (DAC), chip-to-chip and chip-to-module channels.

“Cadence’s latest 112G-ELR IP on TSMC’s N4P process will benefit our mutual customers with significant performance improvement in silicon, helping them address design challenges with the continuous technology advancement from Cadence’s leading IP solutions and TSMC’s advanced process technologies,” said Dan Kochpatcharin, head of the Design Infrastructure Management Division at TSMC. “Our latest collaboration with Cadence promotes the development of new technologies for hyperscale, AI/ML, 5G infrastructure and other applications.”

www.cadence.com/go/112gelr


A10 Networks adds next-gen Web Application Firewall powered by Fastly

A10 Networks introduced a combined solution of its Thunder Application Delivery Controller (ADC) and a new A10 Next-Generation Web Application Firewall (WAF), powered by Fastly, to enable automated, multi-layered security, and resilience.

A10 said its integrated A10 Next-Gen WAF provides deep web application security services. As the solution sits at the primary application ingress point, a single deployment can efficiently front-end one, hundreds, or thousands of applications without the need for individual server end-point deployments. 

"A10 Networks provides high-performance application delivery and security, and its solution is a natural fit for our next-gen WAF to help provide mutual customers with deep web application security for hybrid cloud environments. This is the first software and hardware application delivery controller implementation for Fastly, which expands our addressable market and provides A10 with the most advanced WAF technology for its customers,” said Emily Friedberg, group vice president, global partnerships at Fastly.

Key capabilities:

  • Layered Defense – Counters modern web threats, and includes OWASP Top 10 mitigation, DDoS protection, authentication, and TLS/SSL decryption.
  • Simplification via Consolidation – Converges under Thunder ADC as a fully integrated single appliance solution that is optimized with advanced load balancing, ADC caching, and the Fastly cloud service; customers have a single point of support with A10.
  • Ease of Use – No learning period for IT teams with near-zero false positives; almost 90 percent of Fastly users run in blocking mode, ensuring only bad traffic is stopped.
  • Lower Costs – Provides superior protection with little or no tuning; uses a combination of thresholding, along with Fastly’s proprietary Network Learning Exchange (NLX) and SmartParse technology, to reduce false positives, for highly effective automated detection and blocking.
  • Reduced Risk – Protects against modern attack vectors like account takeover (ATO), enumeration, and DDoS; integrates with popular DevOps and SIEM tools, making it a great fit for enterprise DevSecOps teams.

“When selecting a partner to deliver superior web application firewall security for our hybrid cloud solutions, Fastly was an obvious choice. It has been a Gartner Peer Insights Customers’ Choice for Cloud Web Application and API Protection (WAAP) for the last five years. Our combined solution will help customers ensure security and resiliency while reducing the operational overhead for security teams,” said Mikko Disini, VP of product line management at A10 Networks.


www.a10networks.com

RSA 2023 Video: What's keeping CISOs up at night?

https://youtu.be/ACp4kaNEi8M

The worries include compliance, brand reputation, and liability. Depending on the industry, there are different concerns. In this video from the 2023 RSA Conference in San Francisco, A10 Networks discusses top concerns of Chief Information Security Officers.

For example, in the gaming industry, DDOS attacks are prevalent, and CISOs need to ensure that they have the latest DDOS technology to keep the game running and prevent the loss of users. In contrast, if a company deals with credit card data or personal identifiable information, they need to worry about web application attacks. The concerns of CISOs depend on the organization and business goals.


Accenture Security links with Google Cloud for AI-powered cyber response

Accenture Security will power its new Managed Extended Detection and Response (MxDR) service with security-specific generative artificial intelligence (AI) from Google Cloud.

Accenture's service provides enterprise customers with crisis management, incident response, and threat intelligence.

“Enterprises are rapidly transforming, and they need cyber resilience at the core of their digital journeys to protect their businesses and grow with trust and confidence,” said Paolo Dal Cin, who leads Accenture Security globally. “We’re bringing the best expertise and capabilities from Accenture Security, Google Cloud and Mandiant together to help our clients achieve the cyber resilience level their businesses need in today’s cyber threat landscape. When powered by generative AI, our integrated cybersecurity services should enable a new level of cyber resilience, with greater scale, speed and effectiveness.”

Sunil Potti, GM and VP of cloud security, Google Cloud, said, “Accenture and Google Cloud share a vision for helping enterprises modernize their security positions by combining the most advanced security technology with the world’s leading security experts. Accenture’s new MxDR service is now powered by Chronicle’s world-class security operations, Mandiant’s leading threat intelligence, and new security-specific generative AI capabilities, which will help businesses defend against the most sophisticated cyber threat actors.”

Highlights:

  • Chronicle Security Operations: Accenture’s new MxDR service utilizes the Chronicle platform to enable its security teams to better detect, investigate, and respond to cyber threats. It does so by helping enterprises ingest and search the massive amounts of network and security telemetry they generate, with near real-time analysis and context on security risks and vulnerabilities.
  • Mandiant Threat Intelligence: Accenture’s new MxDR service, which includes its Cyber Threat Intelligence capability, will also embed the actionable, leading threat intelligence of Mandiant. Mandiant provides near real-time access to context and threat intelligence from the frontlines, with a global team of analysts who operate around-the-clock investigating and responding to the most sophisticated cyberattacks.
  • Security-specific generative AI from Google Cloud: Accenture will be the first company to utilize the Google Cloud Security AI Workbench, an industry-first extensible platform powered by a specialized, security large language model (LLM), Sec-PaLM, that leverages Google’s visibility into the threat landscape and Mandiant’s frontline intelligence on vulnerabilities, malware, threat indicators, and more. By integrating generative AI within the service Accenture aims to significantly accelerate incident detection, analysis, and response, mitigating the impact of security incidents.
  • Location- and industry-specific customization: Accenture and Google Cloud can build and manage customized security programs for organizations with complex environments that require more than turnkey solutions. This approach is well-suited for global organizations that need to protect large attack surface areas and high volumes of assets, or for companies needing to comply with local regulations related to data residency and sovereignty.
  • Managed crisis and incident response: Accenture’s security professionals will partner closely with Mandiant threat experts to help organizations embed resilient cybersecurity protections across their business assets and hybrid workforces.
  • Accenture-owned technology accelerators: Accenture security services include unique technology and an intuitive user experience, which increase threat detection and response capabilities and reduce the time and effort needed to contain vulnerabilities.


Arista brings AI/ML enterprise network identity management

Arista Networks introduced an enterprise network identity management capability that leverages AI/ML to  simplify the secure onboarding and troubleshooting for users and devices and the management of ever-expanding security policies.

Arista Guardian for Network Identity (CV AGNI) is based on Arista’s foundational NetDL architecture and leverages AVA (Autonomous Virtual Assist) for a conversational interface that removes the complexity inherent in managing network identity from a traditional legacy NAC solution. AVA codifies real-world network and security operations expertise and leverages supervised and unsupervised ML models into an ‘Ask AVA’ service, a chat-like interface for configuring, troubleshooting and analyzing enterprise security policies and device onboarding. CV AGNI also adds user context into Arista’s network data lake (NetDL), greatly simplifying the integration of device and user information across Arista’s products and third-party systems.

CloudVision AGNI’s key features include the following:

  • User self-service onboarding for wireless with per-user unique pre-shared keys (UPSK) and 802.1X digital certificates.
  • Certificate management with a cloud-native PKI infrastructure.
  • Enterprise-wide visibility of all connected devices. Devices are discovered, profiled and classified into groups for single-pane-of-glass control.
  • Security policy enforcement that goes beyond the traditional inter-group macro-segmentation and includes intra-group micro-segmentation capabilities when combined with Arista networking platforms through VLANs, ACLs, Unique-PSK and Arista MSS-Group techniques.
  • AI-driven network policy enforcement based on AVA for behavioral anomalies. When a threat is detected by Arista NDR, it will work with CV AGNI to quarantine the device or reduce its level of access.

Keysight makes UK’s First 100Gbps 6G Sub-THz connection

Keysight Technologies,  in collaboration with National Physical Laboratory (NPL) and the University of Surrey, has made the first 6G connection at speeds greater than 100 Gbps over sub-terahertz (THz) frequencies in the U.K.

Keysight, NPL, and the University of Surrey established the first sub-THz high throughput 6G testbed in the U.K. to address these challenges. 

Located at NPL, this new 6G testbed achieved the U.K.’s first high-speed sub-THz data link. The demonstration was made at a frequency of 300 GHz using both 32 and 64 quadrature amplitude modulation (QAM). Built on Keysight’s 6G Sub-Terahertz R&D Testbed, the testbed uses the M8194A Arbitrary Waveform Generator (AWG) combined with Virginia Diode Inc. (VDI) upconverters / downconverters to generate the signal and Keysight’s UXR0704A Infiniium multichannel high-performance 70 GHz oscilloscope to analyze the signal.

Irshaad Fatadin, Principal Scientist, National Physical Laboratory, said: “6G is a key focus for NPL and we are using our scientific and measurement capabilities to tackle the challenges of this new technology. Our partnership with Keysight will be a critical success factor in our 6G research work.”

Mosaab Abughalib, Senior Research Director and General Manager for Keysight’s Network Emulation Group, said: “Through this partnership we are bringing Keysight solutions and experts together with scientists from NPL and the University of Surrey to unlock the true potential of 6G.”

https://www.keysight.com/us/en/about/newsroom/news-releases/2023/0424-pr23-070-keysight-makes-uk-s-first-100gbps-6g-sub-thz-conne.html

Sunday, April 23, 2023

CoreWeave raises $221 million for GPU cloud infrastructure

CoreWeave secured $221 million in Series B funding for its cloud infrastructure optimized for large-scale GPU-accelerated workloads. Target workloads include artificial intelligence and machine learning, visual effects and rendering, batch processing and pixel streaming.

CoreWeave said its goal is to offer purpose-built, customized solutions that can outperform larger, more generalized cloud providers. The new capital will also support U.S.-based data center expansion with the opening of two new centers this year, bringing CoreWeave’s total North American-based data centers to five.

“CoreWeave is uniquely positioned to power the seemingly overnight boom in AI technology with our ability to innovate and iterate more quickly than the hyperscalers,” said CoreWeave CEO and co-founder Michael Intrator. “Magnetar’s strong, continued partnership and financial support as lead investor in this Series B round ensures we can maintain that momentum without skipping a beat. Additionally, we’re thrilled to expand our collaboration with the team at NVIDIA. NVIDIA consistently pushes the boundaries of what’s possible in the field of technology, and their vision and guidance will be invaluable as we continue to scale our organization.”

The round was led by Magnetar Capital  with contributions from NVIDIA, and rounded out by Nat Friedman and Daniel Gross.

NVIDIA recently released the highest-performance data center GPU, the NVIDIA H100 Tensor Core, along with the NVIDIA HGX H100 platform. CoreWeave announced at the NVIDIA GTC conference in March that its HGX H100 clusters are live and currently serving clients such as Anlatan, the creators of NovelAI. In addition to HGX H100, CoreWeave offers more than 11 NVIDIA GPU SKUs, interconnected with the NVIDIA Quantum InfiniBand in-network computing platform, which are available to clients on demand and via reserved instance contracts.

https://www.coreweave.com 

Linux Foundation's DentOS 3.0 targets distributed enterprise edge

The Linux Foundation announced the release of DentOS 3.0, code-named "Cynthia," an open source network operating system utilizing the Linux Kernel, Switchdev, and other Linux based projects, including SwitchDev, POE+ and others.

Specific feature updates in 3.0 include:

  • Traffic Control (TC) Persistence
  • Enables traffic control commands and configuration data across boots simplifies configuration and set up
  • New Kernel 5.15(LTS)
  • Increased security, management with new Kernel software
  • Rapid DevOps, providing early access to:
  • IEEE 802.1x :security/patches in kernel and Switchdev
  • QoS: (Mgmt): enables prioritization and optimization of bandwidth usage in remote locations
  • IPv6: for continued expansion and support of more IOT Devices
  • IGMP Snooping: no router required at enterprise location
  • Egress Policer

DentOS enables Amazon's Just Walk Out Technology to connect and manage thousands of devices like cameras, sensors, entry and exit gates, and access points on the network edge. 

"We are pleased to leverage the DENT open-source platforms to power networking infrastructure to enable customers to skip check-out lines with our Just Walk Out Technology," said Jason Long, head of Networking for Amazon Physical Retail Technology and chairman of the DENT Board. "DENT enabled us to reduce our networking costs by giving us access to open-source switches that allowed Amazon to efficiently deploy new hardware and software whenever we need instead of waiting for a bug fix from a third-party vendor".

"Adoption and deployment by the world's largest e-commerce leader with its Just Walk Out Technology is a shining example of the power of open source," said Arpit Joshipura, general manager, Networking, Edge and IoT, the Linux Foundation. "In just three years, the DENT community created a working platform for disaggregated networks to power multiple device locations at the edge, now used by top retailers to streamline operations. This undertaking is only possible by the power of collaborative open source development."  

https://github.com/dentproject/dentOS


New Middle Mile: How does Graphiant overcome network inflexibility?

Ali Shaikh, Chief Product Officer from Graphiant, explains how their network architecture delivers a simple network service. Key takeaways:

Graphiant’s stateless core philosophy ensures that the transit is pure transport, fast and flexible, instead of today’s inflexible and rigid topologies

* Graphiant’s architecture puts metadata into the data plane itself, allowing customer traffic to go wherever it needs to without having to worry about how to manage a complicated customer configuration landscape

* Graphiant’s service is built in such a way that it can become an open standard and shared with the industry

Check out the rest of the New Middle Mile (#nmm2023) video showcase here: https://ngi.fyi/nmm23yt

Want to be in one of NextGenInfra.io videos? Contact us at info@nextgeninfra.io.

Friday, April 21, 2023

QuSecure collaborates with Red Hat on post-quantum cryptography

QuSecure, a start-up based in San Mateo, California, announced that its post-quantum cybersecurity technology is supported on Red Hat Enterprise Linux, Red Hat OpenShift and Red Hat Ansible Automation Platform. 

The QuSecure solution enables organizations to address crypto modernization efforts that government and private enterprises are undertaking to implement a zero-trust quantum resilient architecture before quantum computers begin decrypting today’s data.

“This comprehensive security solution brings together the best of both worlds: QuSecure’s advanced post-quantum cybersecurity technology supported on Red Hat open-source platforms. With this collaboration, we aim to empower organizations of all sizes with the tools they need to safeguard their digital assets against modern cyber threats,” said Red Hat’s Anna Levine, Senior Director, Public Sector Sales, Defense and National Security Programs. 

“After years of discussion, the time has finally come for enterprise and government entities to take action and remediate their systems against the imminent quantum threat,” said Dr. Garrison Buss, Chief Strategy Officer at QuSecure. “We applaud Red Hat for working so closely with us to empower these stakeholders to complete a low-risk upgrade to their currently vulnerable networks with the latest in modern cryptography, breathing new life into their existing infrastructure and allowing them to implement security measures to help protect their systems for the next decade and beyond.” 

https://www.qusecure.com

Quantum Source raises $12 million for photonic quantum computing

Quantum Source (QS), a start-up located in Rehovot Israel, has raised a $12M seed extension investment for its work in photonic computing, which uses photons as a representation of qubits.

Quantum Source, which was founded in 2021, aims to build fault-tolerant quantum systems that scale to millions of qubits and that will have the potential to unleash dramatic acceleration in numerous cutting edge fields that include drug design, material development, cybersecurity, and the processing of large datasets for AI applications.

“We founded Quantum Source with the belief that photonic quantum technologies are the best route to achieve large-scale, fault-tolerant quantum computers. Our unique approach will dramatically improve the scalability of those machines and will be the key to commercial success of quantum computers,” said Oded Melamed, co-founder and CEO of Quantum Source. “Having investors such as Dell Technologies Capital believe in us will allow us to accelerate our work and by extension, entire industries.”

Quantum Source is led by four co-founders: CEO Oded Melamed; VP of R&D, Gil Semo; Chief Scientist Prof. Barak Dayan; and Chairman Dan Charash. The senior leadership has successfully co-founded, scaled, and sold deeply technical companies to Apple, Broadcom, and Sony. With more than 170 articles published among them, the broader team is composed of 18 PhDs in either physics or electrical engineering from esteemed universities including Caltech, Columbia University, MIT, Weismann Institute of Science, and Yale University.

The funding round was led by Dell Technologies Capital with participation from 10D as well as existing investors Eclipse VC, Grove Ventures, and Pitango First. The extension brings Quantum Source’s seed round total to $27 million.

http://www.qs-labs.com


Thursday, April 20, 2023

Broadcom delivers Jericho3-AI switching fabric

Broadcom began shipping its Jericho3-AI switching silicon chip for delivering high-performance Ethernet for GPU clusters supporting AI workloads.

The Jericho3-AI fabric offers 26 petabits per second of Ethernet bandwidth, almost four times the bandwidth of the previous generation, while simultaneously delivering 40 percent lower power per gigabit. The chip boasts advanced capabilities such as load balancing, congestion-free operation, ultra-high radix, and zero-impact failover.

Broadcom said it optimized the chip to handle the unique characteristics of AI workloads such as a low number of large, long-lived flows, all starting concurrently upon completion of an AI computation cycle. 

“The benchmark for AI networking is reducing the time and effort it takes to complete the training and inference of large-scale AI models,” said Ram Velaga, senior vice president and general manager, Core Switching Group, Broadcom. “Jericho3-AI delivers significant reduction in job completion time compared to any other alternative in the market.”

Highlights

  • "Perfect load balancing" equally sprays traffic over all links of the fabric, ensuring maximum network utilization under the highest network loads.
  • Congestion-free operation with end-to-end traffic scheduling ensures no flow collisions and no jitter.
  • Ultra-high radix uniquely allows the Jericho3-AI fabric to scale connectivity to 32,000 GPUs, each with 800Gbps, in a single cluster.
  • Zero-Impact Failover functionality ensures sub-10ns automatic path convergence, resulting in no impact to job completion time.
  • Long-reach SerDes, distributed buffering, and advanced telemetry, all provided using industry-standard Ethernet. 

“Cloud operators will upgrade their AI infrastructure to address the massive growth in bandwidth, driven by a new generation of high-capacity GPUs and the emergence of large language models,” said Bob Wheeler, principal analyst at Wheeler’s Network. “Jericho3-AI offers a high-bandwidth, low-latency and low-power choice for networks connecting tens of thousands of GPUs, revolutionizing the economics of building and maintaining AI clusters for this exciting new era.”

https://www.broadcom.com/company/news/product-releases/61156

OCP releases chiplet interconnect link layer spec and BoW 2.0

The Open Compute Project Foundation (OCP) released of a new chiplet interconnect link layer specification and an evolution of its previously-released Bunch of Wires (BoW) PHY specification. 

  • The development of the OCP ODSA (Open Domain-Specific Architecture) Link Layer specification is focused on the silicon die disaggregation use case maintaining simplicity to serve as many markets as possible, with key features, such as: (1) extensibility defining interface profiles allowing the addition of new link layer features, customization to bus protocols and enabling Chiplet interconnect interoperability, (2) portability across different die implementation methodologies and process nodes, (3) ability to scale with support for multiple PHY slices and different data rates, (4) while keeping latency low, especially with use of Forward Error Correction (FEC) eliminating where possible serialization due to Cyclic Redundancy Check (CRC).
  • BoW 2.0 includes significant advancements on top of doubling top speed to 512 Gbps per 16 lane slice, including: (1) new energy efficient operations with gated clock and data line inactive modes that can save up to 90% of power consumed for interconnect, (2) half-width slices for dies with smaller absolute beachfront requirements, (3) sideband slices for channel management, (4) bidirectional slices with  direction configurable across channel, and (5) and bump redundancy for improved reliability on failure.

“The OCP recognized several years ago that innovation in silicon needed to be amplified, just as the constraints of larger silicon dies were beginning to impede progress. The OCP created its vision for an open Chiplet economy with the intent of being a catalyst for change by establishing a strong community to deliver on the new standardizations, tools and best practices around technical and business workflows that would be required for a truly open economy, where vendors would sell Chiplets embodying their IP to integrators that would build specialized System in Package (SiP)s. Moving forward the OCP intends to become the front door to an open Chiplet marketplace, providing standardizations allowing Chiplet vendors to self-certify, or 3rd party organizations to independently verify, Chiplet Known Good Die (KGD) characteristics," said Cliff Grossner Ph.D., VP Market Intelligence and Innovation at the Open Compute Project Foundation.

“We can see the first visible seedlings of the open Chiplet economy sprout with the release of LL 1.0 and BoW 2.0 complementing the recent release of CDXML and establishing an alliance with JEDEC and work on standardized Chiplet test benches and interconnect benchmarks. We are at an inflection point where Chiplet vendors have a fairly complete set of standards and tools available to derisk building Chiplet product sets. The market leaders developing Chiplets are still playing a very important role filling gaps that they identify in current standards and we expect to see these additional learnings materialize as updates to existing standards as the open Chiplet economy matures,” said Bapi Vinnakota Ph.D., ODSA Project lead at OCP.

https://www.opencompute.org/blog/open-compute-project-pushes-fast-forward-on-an-open-chiplet-economy



AWS contributes New Edge Gateway spec to OCP

AWS has contributed a new edge gateway base specification to the Open Compute Project 

The spec is targeted for deployments supporting cloud-delivered managed edge services in various market segments, including retail. The edge gateway will also support OCP’s Switch Abstraction Layer (SAI) and thus end users will have a choice of Network OS including SONiC or DENT from the Linux Foundation. 

The specification allows for gateways configured with 48 ports (32x1Gb ports and 16x2.5Gb ports) or 24 ports (24x2.5 Gb ports), both configurations include 4x25Gb SFP28 Uplink Ports. The gateway can support PoE (15.4 to 100 Watts) and optionally Cellular LTE and/or 5G and/or Wi-Fi.  The gateway CPU can be a pluggable module with extended computational performance to allow for demanding workloads of SD-WAN and SASE. Software support includes the ONIE loader and SAI providing choice of SONiC or DENT as Network OS. Hardware management leverages OCP Hardware Baseline Management Profile. For security, TPM 2.0 Module, Secure Boot, and MACSec are supported.

OCP envisions tat this gateway specification will be a catalyst for development of a new and resilient open enterprise edge computing supply chain, with participation from ODM and OEM network equipment vendors including Edgecore, Celestica, Delta Networks, Extreme Networks, Ragile Networks, Delta Electronics and Wistron. The expectation is that production prototypes will be available for display at the OCP Global Summit in October 2023, and that volume shipments will begin in 1Q 2024.

“Collaboration across open ecosystems is key to digital transformation,” said Arpit Joshipura, general manager, Networking, Edge, and IoT, the Linux Foundation. “Integrating with other thought leaders like OCP has helped the development of our open source NOSes, DENT and SONiC, especially in scaling and securing enterprise edge deployments across use cases. We are especially eager for new developments and specifications harmonizing SwitchDev and SAI for more intuitive deployment.”

https://www.opencompute.org/blog/open-compute-project-foundation-expands-open-edge-ecosystem

Zayo expands LH dark fiber and 400G routes, upgrades IP core with Juniper

Zayo announced a series of expansions and enhancements to its network and services, including significant growth of its long-haul dark fiber and 400G-enabled routes and modernization of its IP core network. Juniper was selected for the IP core upgrade.

“The only way to stay ahead of the digital curve is to continuously transform. Transformative ideas need a reliable, resilient and on-demand network,” said Bill Long, Chief Product Officer at Zayo. “Zayo is leading the industry with network automation and self-service options, ensuring customers have unprecedented speed and resilience with more flexibility and elasticity, while enhancing security and value, so our customers can focus on making progress toward their business goals instead of worrying about their network.”

Zayo is launching Waves on Demand to enable same-day turn-up on the most in-demand routes, with significantly shortened delivery times. Customers can quickly provision “Wavelength on Demand” between key data center locations across its market-leading network footprint, including its highest-demand routes. In 2023, Zayo launched 8 new Waves on Demand routes, with 5 additional routes planned for the future.


Zayo’s Completed Waves on Demand routes include:

  • Newark, NJ - New York, NY (up to 400G)
  • Ashburn, VA - New York, NY
  • Hillsboro, OR - Seattle, WA
  • Ashburn, VA - Newark, NJ
  • Atlanta, GA - Dallas, TX
  • Los Angeles, CA - San Jose, CA
  • Inter-Los Angeles, CA
  • Los Angeles, CA - San Jose, CA (alt)

Zayo’s Planned Waves on Demand Routes include:

  • Toronto, ON - Chicago, IL
  • San Jose, CA - Seattle, WA
  • Newark, NJ - Chicago, IL
  • Chicago, IL - Secaucus, NJ
  • Englewood, FL - Chicago, IL

This year Zayo began IP Core upgrades to support 400G connectivity. 

In 2022, Zayo added 5,200 route miles to its network, resulting in more than 1.35M fiber miles.

Zayo now has 224 400G-enabled wavelength points of presence (PoPs) and 145 100G-enabled PoPs.

Zayo deployed 24 long-haul waves routes in 2022 with 926TB of wavelength capacity, enabling 400G services across these routes, spanning more than 20,000 route miles. In 2023, Zayo will exceed the number of new Long Haul Dark Fiber routes deployed in 2022.

Zayo will complete 8 long-haul construction projects in 2023, totaling 2,951 route miles and 708,000 fiber miles.

Zayo is estimated to complete 32 400G routes in 2023 with 14 completed in the first half of the year.


Zayo’s 2023 planned new and augmented dark fiber routes:

  • New - St. Louis, MO to Indianapolis, IN
  • Overbuild - Denver, CO to Dallas, TX
  • Overbuild - Chicago, IL to Omaha, NE
  • Overbuild - Omaha, NE to Denver, CO
  • Overbuild - Seattle, WA to Vancouver, WA
  • New - Columbus, OH to Pittsburgh, PA
  • Overbuild - St. Louis, MO to Memphis, TN
  • New - Columbus, OH to Ashburn, VA

Zayo’s New Tier 1 400G Routes:

  • Albany, NY - Newark, NJ
  • Bend, OR - Umatilla, OR
  • Chicago, IL - Cleveland, OH
  • Albany, NY - Boston, MA
  • Atlanta, GA - Washington, DC
  • Dallas, TX - St. Louis, MO
  • Denver, CO - Dallas, TX
  • Kansas City, MO - Indianapolis, IN
  • Las Vegas, NV - Phoenix, AZ
  • Montreal, QC (Canada) - Quebec City, QC (Canada)
  • Columbus, OH - Ashburn, VA
  • Columbus, OH - Cleveland, OH
  • Columbus, OH - Pittsburg, PA
  • Chicago, IL - Clinton, KY
  • Clinton, KY - Ponchatoula, LA
  • Toronto, ON (Canada) - Waterloo, ON (Canada) (Crosslake)
  • Toronto, ON (Canada) - Montreal, QC (South) (Canada)
  • Toronto, ON (Canada) - Montreal, QC (North) (Canada)
  • Indianapolis, IN - Columbus, OH
  • Ashburn, VA - Baltimore, MD
  • Salt Lake City, UT - Seattle, WA
  • Los Angeles, CA - San Jose, CA

On the security front, Zayo is deploying Resource Public Key Infrastructure (RPKI) filtering - a component of Mutually Agreed Norms for Routing Security (MANRS) compliance designed to secure the internet’s routing infrastructure - Zayo now requires two-factor authentication process for Border Gateway Protocol (BGP) route management. 

https://www.zayo.com/newsroom/zayo-unveils-significant-network-expansion-and-industry-first-product-innovation-to-enable-customers-to-connect-whats-next/


stc's TAWAL acquires telecom towers in Bulgaria, Croatia, Slovenia

stc Group's infrastructure subsidiary, TAWAL, agreed to acquire United Group’s telecommunications tower assets fir EUR 1.22 billion.

Following completion of the acquisition, TAWAL will own and operate more than 4,800 sites across Bulgaria, Croatia, and Slovenia (all European Union member states, two of which are already members of the Eurozone), providing the full range of passive infrastructure services ranging from ground-based towers, rooftops small cells to in-building-solutions. As part of the 20-year master services agreement with United Group, TAWAL will deploy over 2,000 additional sharable sites, while co-location relationships with other mobile network operators will be maintained and expanded, enabling stc Group to drive digital transformation through providing world-class connectivity.

This marks the Saudi Arabian telecoms group's first investment in the European market. The company says the step marks a major milestone in its international expansion journey. TAWAL currently owns a portfolio of over 16,000 telecom towers. The company is actively supporting digital transformation plans in Saudi Arabia, expanding its reach across new cities and rural areas in the Kingdom and actively rolling out smart-city-ready technologies such as camouflage telecom towers, smart poles capable of hosting 5G and IoT applications, in-building solutions, and small cells.

Olayan Alwetaid, Chief Executive Officer, stc Group, said: “Our agreement with United Group represents an exciting new chapter for TAWAL and the wider stc Group. The agreement is a significant milestone in our ambitious growth strategy and the expansion of our international footprint"


AT&T posts flat results for Q1, up 1% yoy

AT&T reported Q1 communications revenue of $29.2 billion, up 1.0% year over year due to increases in Mobility and Consumer Wireline, which more than offset a decline in Business Wireline. Operating income was $6.7 billion, up 3.9% year over year, with operating income margin of 23.1%, compared to 22.5% in the year-ago quarter.

"Our teams take pride in connecting more people to greater possibility through 5G and fiber,” said John Stankey, AT&T CEO. “We’re winning thanks to a proven and sustainable playbook that centers on simple, customer-centric experiences. As a result, we’re adding high-value customers, and when they choose AT&T, they stay with us. The work we’re doing today is establishing a foundation for durable, long-term growth, and we remain confident in our full-year guidance.”

Some highlights:

  • Total wireless net adds were 5.1 million including:
  • 542,000 postpaid net adds with:
  • 424,000 postpaid phone net adds
  • (56,000) postpaid tablet and other branded computing device net losses
  • 174,000 other net adds
  • 40,000 prepaid phone net adds
  • Postpaid churn was 0.99% versus 0.94% in the year-ago quarter.
  • Postpaid phone churn was 0.81% versus 0.79% in the year-ago quarter.
  • Prepaid churn was 2.73%, with Cricket substantially lower, versus 2.77% in the year-ago quarter.
  • Postpaid phone ARPU was $55.05, up nearly 2.0% versus the year-ago quarter, due to prior-year pricing actions, higher international roaming and a mix shift to higher-priced unlimited plans.
  • FirstNet connections reached approximately 4.7 million across more than 25,000 agencies.
  • More than 750,000 U.S. business buildings are lit with fiber from AT&T and more than 10 million business customer locations are on or within 1,000 feet of AT&T fiber.
  • Total broadband net losses, excluding DSL, were 23,000, reflecting AT&T Fiber net adds of 272,000, more than offset by losses in non-fiber services. 
  • AT&T Fiber now has the ability to serve 19.7 million customer locations and offers symmetrical, multi-gig speeds across parts of its entire footprint of more than 100 metro areas.
  • In Mexico, total wireless net adds were 10,000, including 58,000 prepaid net losses, 49,000 postpaid net adds and 19,000 reseller net adds.

Leonardo migrates its defense IT systems to Azure

Leonardo, a defense contractor known for its helicopters and advanced electronics, has deployed the cloud-based Azure platform across its UK business.

Leonardo’s UK-based scientists and engineers will now have secure access to a remotely-accessible ‘digital backbone’ that will speed up product research, cut development costs and enable closer collaboration with customers and other partners. In the future, Leonardo will also be able to use the digital backbone to exploit the huge amount of data it collects in the form of new products and services.

Leonardo said it will invest around £100M this year in the UK in its ongoing enablers strategy, including the Future Factory transformation project. Last year the company rolled out digital electronics factories across the country, with a ‘common data environment’ putting big data at the fingertips of engineers. 

Gareth Hetheridge, Director Digital and IT, Leonardo UK, said: “With the introduction of the Leonardo digital backbone, employees, partners and customers can now securely access relevant data and applications anytime, anywhere. This will be vital as industry moves to a model of defence contracting where companies and customers work more closely together than ever before, despite being physically located around the country and internationally. Our close partnership with Microsoft UK and Accenture has allowed us to adopt this disruptive technology at pace and we’re very proud to be flying the flag for bringing the benefits of the cloud to the UK defence industry.”

Aaron Neil, Director Defence & Secure Markets, Microsoft UK, said: “Over the last year we have worked with Leonardo UK to better understand its digital transformation aspirations and priorities. We are delighted to support Leonardo's migration to Azure, enabling highly secure access to the cloud; this announcement is testament to our close partnership and an exciting milestone in delivering advanced technology to accelerate capability delivery for UK Defence.”



Wednesday, April 19, 2023

Marvell advances its high-speed interconnects on TSMC's 3nm node

Marvell demonstrated high-speed, ultra-high bandwidth silicon interconnects produced on TSMC's 3-nanometer (3nm) process. This includes Marvell's 112G XSR SerDes (serializer/de-serializer), Long Reach SerDes, PCIe Gen 6 / CXL 3.0 SerDes, and a 240 Tbps parallel die-to-die interconnect.

SerDes and parallel interconnects serve as high-speed pathways for exchanging data between chips or silicon components inside chiplets. Together with 2.5D and 3D packaging, these technologies will eliminate system-level bottlenecks to advance the most complex semiconductor designs. SerDes also help reduce pins, traces and circuit board space to reduce cost. A rack in a hyperscale data center might contain tens of thousands of SerDes links.

The new parallel die-to-die interconnect, for example, enables aggregate data transfers up to 240 Tbps, 45% faster than available alternatives for multichip packaging applications. 

Marvell incorporates its SerDes and interconnect technologies into its flagship silicon solutions including Teralynx switches, PAM4 and coherent DSPs, Alaska Ethernet physical layer (PHY) devices, OCTEON processors, Bravera storage controllers, Brightlane™ automotive Ethernet chipsets, and custom ASICs. Moving to a 3nm process enables engineers to lower the cost and power consumption of chips and computing systems while maintaining signal integrity and performance.

"Interconnects are taking on heightened importance as clouds and other computing systems grow in size, complexity and capability. Our advanced SerDes and parallel interfaces will play a significant role in providing a platform for developing chips with best-in-class bandwidth, latency, bit error rate, and power efficiency for meeting the demands of AI and other complex workloads," said Raghib Hussain, president of products and technologies at Marvell. "We are proud to be able to deliver such advances on TSMC's 3nm technology and take semiconductor designs to the next level for our customers around the world."


  • Marvell was also the first data infrastructure silicon supplier to respectively sample and commercially release the 112G SerDes and has been a leader in data infrastructure products based on TSMC's 5nm process.

https://www.marvell.com/company/newsroom/marvell-demonstrates-industrys-first-3nm-data-infrastructure-silicon.html