Tuesday, March 30, 2021

Cisco unveils Routed Optical Networking

Cisco announced a number of  Internet infrastructure products and technologies including a new Routed Optical Networking solution featuring Acacia pluggable optics and Cisco’s full mass-scale routing portfolio. The company claims its integration of routing and optics will lead to savings of up to 46% over traditional, layered IP + optical product architecture.

At its virtual Cisco Live! event, the company said Acacia’s pluggable coherent optics, advancements in Segment Routing and Ethernet VPN, and new Cisco Crosswork Cloud capabilities will lead to greater efficiency for service providers and web scale companies such as Airtel, Altibox, Eolo, Facebook, Google Cloud, Rakuten Mobile, SFR, Telenor, Telia Carrier, Telstra, Websprix, etc.

Here are some highlights of the announcements

Cisco Silicon One - Cisco first introduced its programmable networking silicon architecture in 2019. The company has now expanded the Cisco Silicon One platform from a routing-focused solution to one which also addresses the web-scale switching market, offering ten networking chips (devices) ranging from 3.2 Tbps to 25.6 Tbps.

Subscriber Management - Cisco is introducing its new Cisco Cloud Native Broadband Network Gateway for telco customers (wireline), joining the Cisco family of existing cloud native broadband routers for cable and mobile. It paves the way for convergence to a unified subscriber management solution, bringing further simplification and efficiency while enabling service providers to offer truly access-agnostic services independent from where people use these services.  

Routing for the Access, Aggregation, Edge, and Core 

  • Cisco’s latest 8000 family of routers now features Cisco Silicon One Q200 series chips offering up to 14.4 Tbps total capacity, enabling 32 and 64 x 100G web-scale switches.
  • Introducing powerful new line cards and chassis for the Cisco Aggregated Service Router (ASR) 9000 series and Network Convergence System (NCS) 500 and 5500 series routers providing increased capacity with capital and operational cost savings
  • New Crosswork Network Controller (CNC) features help customers operate the Cisco Routed Optical Networking solution

Cisco Crosswork Cloud - new application called Traffic Analysis offers a comprehensive view across network peering points. With this insight

Cisco Business Critical Services - design and consulting services to help customers adopt Cisco's Routed Optical Networking and Cloud Native Broadband solutions. 

“Cisco has spent the last five years researching and investing in this portfolio of innovation, focusing on how to help our customers deliver the best internet while being able to grow revenue, reduce their costs and mitigate risk,” said Jonathan Davidson, Senior Vice President and General Manager, Mass-Scale Infrastructure Group, Cisco. “By helping our customers make the right decisions for their networks today, we are setting the world up for success, to connect more people, places and things than ever before. We can all look back on this point in time in the next ten years and celebrate how we rose to the challenge and did the right thing to take care of the internet.”


Fungible unveils its multi-tenant data center platform for pooled resources

Fungible, a start-up based in Santa Clara, California, unveiled its multi-tenant, secure data center solution based on pooled resources for enabling the cost-efficiencies of hyperscale data center operators. 

Fungible, which was co-founded by Pradeep Sindhu, the founder of Juniper Networks, and Bertrand Serlet, who is perhaps best known for leading the development of Mac OS X and IoS, has as its fundamental building block a new type of processor called a data processing unit, or DPU -- the so-called “third socket” about the CPU and GPU. So far, Fungible has introduced two versions of its DPU:

  • Fungible F1 DPU – an 800Gbps processor for high-performance storage, analytics and security platforms, and the
  • Fungible S1 DPU – a 200Gbps processor optimized for host-side use cases including bare metal virtualization, storage initiator, NFVi/VNF applications and distributed node security.

Its first shipping product, which was introduced in October 2020, is an NVMe Storage Cluster in a 2RU form factor, and scaling linearly to 300M IOPS in a single 40RU rack.  The company says its high-performance design improves $/IOPS significantly by consolidating workloads and increasing utilization of storage media.

The newly-announced Fungible Data Center solution extends this vision to power hyper-disaggregated, composable infrastructure with bare-metal performance for multiple tenants. The solution is orchestrated by Fungible Data Center Composer software, which was developed internally, accelerated through the recent acquisition of Cloudistics. This is accomplished via a single pane of glass management for dynamic workload changes with bare metal performance. Security is maintained with independent hardware-accelerated security domains, fine-grained segmentation, robust QoS and line-rate encryption.

Key elements of the Fungible Data Center include:

  • Standard compute and GPU servers equipped with the Fungible Data ServicesPlatform - a standard full-height, half-length PCIe card powered by a Fungible S1 DPU. The Fungible Data Services Platform card comes at three performance points: 200G, 100G and 50G
  • Fungible Storage Cluster comprising a cluster of Fungible FS1600 scale-out disaggregated storage nodes, each powered by two Fungible F1 DPUs
  • Standard TOR switches and routers for data, BMC and management
  • Fungible Data Center Composer - a centralized software suite that enables bare metal composition, provisioning, management and orchestration of infrastructure at all scales

“At Fungible, we believe that if we build a solution that addresses the most challenging requirements in data centers - specifically, hyperscale data centers running the most data-intensive applications, then data centers of all scales, on-premise or cloud, core to edge will reap the benefits as well,” said Pradeep Sindhu, CEO and Co-Founder of Fungible. “Leveraging the Fungible Data Processing Unit™ (DPU), we were able to ‘hyperdisaggregate’ compute and storage resources providing performance, scale and cost efficiencies not even achievable by hyperscalers. Today, we deliver the ‘composable’ piece with the first incarnation of Fungible Data Centers, fully managed by the innovative Fungible Data Center Composer software.”

“The Fungible Data Center Composer is developed by fully harnessing the powerful capabilities of the Fungible DPU,” said Dr. Srinidhi Varadarajan, Senior Vice President of Solutions at Fungible. “The result is a data center architecture that is simple, flexible, high performance and secure. We believe the Fungible Data Center is the inevitable way to build infrastructure because it dynamically adapts to changing workload conditions, while delivering performance equivalent to that of a custom designed bare metal data center per workload.”

“We’ve identified the bottlenecks within our existing distributed infrastructure architectures. To achieve our desired system efficiency targets we are looking to new hardware approaches that better support our infrastructure processing stacks,” said Brad Settlemyer, Sr. Scientist in Los Alamos’ HPC Design group. “The co-design ofsoftware and hardware to support data services and data analysis is integral to meeting our efficiency targets and advancing our national security mission.”

Fungible is also in strategic partnerships with key OEMs to deliver first-class worldwide deployment, support and training, including Supermicro and Juniper Networks.

“Juniper Networks exists to provide an experience-first network for our customers and to simplify the complexities of modern networking so organizations of all sizes can focus on innovating and moving their businesses forward,” said Raj Yavatkar, CTO, Juniper Networks. “We are excited to be working with Fungible to deliver the Fungible Data Center solution. The combination of Juniper and Fungible technology enables consistent quality of service and the high performance required of today’s cloud dat centers without creating more workflows and layers of technology to manage.

Fungible Data Centers - a deep dive discussion

The networking industry has been on an evolutionary path from monolithic systems toward open, disaggregated and virtualized infrastructure.  Here, we feature a deep dive discussion on the Fungible Data Center architecture with Srinidhi Varadarajan, SVP Solutions at Fungible, and Pardha Pyla, VP Product Solutions & Design at Fungible.


Q&A Topics

04:33 - Fungible has a very audacious vision. How are you doing this?

06:22 - There is no shortage of innovation for data center software, but Fungible’s starting point is its DPU. What is special about it?

08:31 - Where does the DPU fit?

11:02 - Is True Fabric proprietary? Do I need special switches to make it work? Why TCP?

12:17 - What exactly are you announcing?

12:46 - What problems does the Fungible Data Center solve?

15:10 - How does the FDC address these challenges?

16:31 - Not everyone is familiar with bare metal composition. Can you elaborate on how that is different from virtualization? 

18:17 - Do you consider the FDC a form or composable, disaggregated architecture?

22:25 - Are the benefits of the FDC similar to virtualization? 

23:42 - Who are the target customers for the Fungible Data Center?

24:44 - How does the FDC enter the market without having customers start from scratch? What is the insertion point for cloud data centers or telco data centers?

27:13 - Are there any cloud or telco workloads in particular that are especially well-suited for this architecture?

28:05 - What is the insertion point opportunity for enterprise data centers?

31:05 - How does Fungible address the high-performance compute (HPC) opportunity? What is the insertion point for AI and ML data center architecture?

 32:53 -  Let’s talk about the software side of the FDC solution. What are the unique innovations behind the Fungible Data Center Composer?

 35:11 - Clearly Fungible is not stopping here. What comes next? What about optimizing GPUs and memory?

ADVA launches DCI open line system designed for 400ZR

ADVA launched a new FSP 3000 data center interconnect (DCI) open line system (OLS) specifically designed for 400ZR technologies.

The ADVA FSP 3000 DCI OLS, which is fully compliant with the OIF’s implementation agreement for 400ZR coherent optical interfaces, isptimized for amplified 400 Gbps DWDM links with reaches up to 120km at 75GHz or 100GHz channel spacing. ADVA says the solution exceeds all 400ZR performance specifications and is also compatible with OpenZR+ interfaces, which expand the application area to DCI applications with distances far beyond 120km. The ADVA FSP 3000 DCI OLS integrates multiple diagnostic tools, including built-in OTDR for the monitoring of all fibers and built-in optical channel monitoring. The solution is currently undergoing customer testing and evaluation.

“Today we’re announcing a major milestone for open, high-capacity data center networks. The launch of our new DCI OLS will be key for operators. It provides the missing link to real-world deployments of open infrastructure built with 400ZR interfaces,” said Christoph Glingener, CTO, ADVA. “Our pedigree in the DCI space is second to none. We were the original DCI company, enabling Fortune 500 companies around the globe to harness the cloud with best-in-breed solutions. For more than 25 years, our innovation has helped the industry evolve. Now we’re giving large-scale DCI providers what they need to move forward. Our next-generation FSP 3000 DCI OLS offers a way to inject ultra-high capacity while reducing total cost of ownership and operational complexity. What’s more, the ADVA FSP 3000 DCI OLS solution is designed to optimize performance with 400ZR-compliant optics from any vendor.”


Innovium offers whitebox certified switches with SONiC software

Innovium is now offering a line-up of TERACertified switch solutions that are pre-validated with optics, DAC and SONiC network operating system software. 

The TERACertified solutions are based on the company's TERALYNX switch silicon and SONiC software. A key advantage of TERACertified is that there is no vendor lock-in, following the same approach as the top few hyperscalers. The solutions are partner switch systems designed in partnership with key ODM partners, delivered with Innovium’s rigorous qualification.  They are ready-to-deploy and built from silicon up, pre-validated with optics, cables and the latest version of SONiC. The portfolio includes 100G, 200G and 400G Ethernet switches with 10/25G to 400G connectivity.The TERACertified lineup includes

  • 1RU fixed switch with 32-ports of QSFP28 (100G) -- a ToR switch in 1RU based on 3.2T TERALYNX 5 switch ASIC
  • 1RU fixed switch with 32-ports of QSFP56 (100/200G) --a  ToR switch in 1RU, based on 6.4T TERALYNX 5 switch ASIC. It is designed for customers as they transition from 25G NRZ to 50G PAM4 connectivity to servers & storage
  • 1RU fixed switch with 32-ports of QSFP-DD (400G) --the highest radix & performance switch in 1RU today, based on 12.8T TERALYNX 7 switch ASIC. This design is targeted at use cases for leaf, spine, core and DCI in the data center and it delivers the most power efficient 12.8T switch with best TCO.
  • 1RU fixed switch with 32-ports of QSFP-DD800 (800G) -- the highest radix & performance switch in 1RU, based on 25.6T TERALYNX 8 switch ASIC. It is targeted at use cases for next-gen leaf, spine, core and DCI in the data center. It delivers the most power efficient 25.6T switch with best TCO

“Our vision is to bring the benefits of open and disaggregated solutions being deployed by top hyperscale customers to the broader market. Using data center optimized TERALYNX switch silicon and SONiC software we are delivering a set of highly robust and validated switches with disruptive TCO,” said Rajiv Khemani, CEO and Co-Founder of Innovium Inc. “Through this effort, we are excited to be transforming the networking industry and ushering in a new era which will deliver breakthrough advantages we have enabled for top hyperscalers to a broad set of customers.”


Imec demos ultrasound sensor in silicon photonics

Imec, a leading research and innovation hub in Belgium, has demonstrated an optomechanical ultrasound sensor on a silicon photonic chip that has an unprecedented sensitivity due to an innovative optomechanical waveguide.

By leveraging this high-sensitivity waveguide, Imec's 20-µm small sensor has a detection limit two orders of magnitudes better than piezoelectric elements of identical size. 

Potential clinical and biomedical applications of ultrasonic and photoacoustic imaging include deep-tissue mammography and the study of vascularization or innervation of potential tumorous tissue. 

 “The sensor we have demonstrated will be a gamechanger for deep tissue imaging in otherwise non-transparent tissues such as skin or brain. For applications such as sub-cutaneous melanoma imaging or mammography, it enables a more detailed view of the tumor and vascularization around, aiding in a more detailed diagnosis,” says Xavier Rottenberg, fellow wave-based sensors and actuators at imec.

Imec said its solution is based on a highly sensitive split-rib optomechanical waveguide fabricated using new CMOS-compatible processing. The sensitivity is two orders of magnitude larger than a state-of-the-art device. A low detection limit can improve the trade-off between imaging resolution and depth for ultrasound applications, and is crucial for photoacoustic imaging, where pressures are up to three orders of magnitude lower than in conventional ultrasound imaging techniques. 

Finally, a fine-pitched (30 µm) matrix of these tiny (20 µm) sensors can be easily integrated on-chip with photonic multiplexers. This opens the possibility of new applications such as miniaturized catheters because the sensor matrices require only few optical fibers to be connected instead of one electrical connection per element in the case of piezoelectric sensors.


Keysight supplies cloud-native 5G core network test solution

NEC has selected Keysight Technologies' scalable and cloud-native 5G core (5GC) network test solution, LoadCore, to validate its 5GC platform.

Keysight’s LoadCore 5GC network test solution enables NEC to perform network capacity tests, measure device data throughput and model a wide variety of complex end-user behavior and traffic mobility scenarios. Next generation communications, based on 5G NR technology, uses wider bandwidths to manage higher data throughput and traffic flows compared to 4G LTE. LoadCore allows NEC to efficiently optimize performance testing of user plane functions deployed in 5G NR networks.

“We’re pleased that NEC, a leading provider of 5G core network functions and radio units, virtualized radio access network software and network transport equipment, selected Keysight to validate the user plane function (UPF) in their 5GC platform,” said Kalyan Sundhar, vice president and general manager for Keysight’s 5G edge to core industry group. “Keysight’s portfolio from the edge of the RAN to the core of the network, including open RAN platforms, enables vendors and mobile operators to verify the interoperability, performance, conformance and security of 5G networks.”

Indosat Ooredoo sells 4,200 towers to Digital Colony for $750 million

PT. Indosat Tbk. (Indosat Ooredoo), a leading service provider in Indonesia, agreed to sell more than 4,200 telecommunications towers to PT EPID Menara Assetco, an Indonesian subsidiary of Edge Point Singapore, which is wholly owned by Digital Colony.

The deal, which includes a lease-back agreement, is valued at USD 750 million, including a supplementary offer, making it one of the largest deals of its kind in Asia.

President Director and Chief Executive Officer of Indosat Ooredoo, Ahmad Al-Neama, said: “I am delighted that Indosat Ooredoo has agreed this deal with Edge Point Indonesia, which has been long in the planning and is in-line with our turnaround strategy to create more value for shareholders and customers. The deal marks the third and final sale of assets from Indosat Ooredoo’s high-quality tower portfolio that moves us towards a more asset-light model and greater focus on delivering outstanding mobile digital services to our customer. I would like to thank the Government of Indonesia for creating a policy environment that has enabled the telecoms industry to thrive and attract high-quality investors such as Edge Point Indonesia. Increasing investment into the sector will ultimately benefit customers and the Indonesian people.”

Suresh Sidhu, CEO of Edge Point Group, welcomed the transaction : "This deal is one of the largest of its kind in Asia and cements our position as a leading tower provider in Indonesia. We are delighted to have further strengthened our strategic partnership with Indosat Ooredoo, which has been developing over several years. The addition of 4,200 towers to our portfolio will help to create significant value for all our stakeholders.”