Contact us
Join Community
Contact us

Exciting Announcement! In celebration of launching our AI Certification, we’re thrilled to offer a 50% discount exclusively. Seize this unique chance—don’t let it slip by!

Learn more

Author: Nisha Syed Ali

Categories
All

Everything You Need to Know About Network Link Optics

Ethernet cables connected to network switch, close up

Optical communication networks are, literally, the backbones of the information superhighway. They have been providing the conduits over which broadband data is delivered worldwide, at the actual speed of light. Hundreds of millions of miles of deployed optical fiber interconnect continents, nations, cities, neighborhoods, and private homes. Behind this backbone comes a suite of complementary optical subsystem components that are pivotal to the operation and management of these networks. These optical microsystems directly interact with the optical signal and – through functionality afforded by design – are able to filter, switch, attenuate, and adapt the optical communication channels carried by the network.

One of the major optical microsystem applications is the optical transceiver. In simple words, an optical transceiver is a device that converts data to light and vice versa to be transmitted and received across optical fibers. Imagine these devices to be traffic merge points in an interstate highway system that takes in off-highway city traffic and links them into super-speed highway traffic. Just as the proper functioning of traffic merges is necessary for the seamless functioning of a highway system, continuous observability of thousands of these devices at scale becomes a must-have for network administrators.

In this blog, we will demystify optical transceiver-based link optics from an engineering perspective. Besides that, we will also highlight the ways Aviz ONES is leveraged by network teams to continuously observe, investigate and triage in real-time, thereby alleviating their time, cost, and resource burdens.

Optical Transceiver – Engineer’s Definition

‍An Optical transceiver is used to convert electrical signals to optical light signals and optical signals to electrical signals. It is a hot-swappable device that can be plugged into a networking device that can send and receive data. Optical transceivers come in different forms and dimensions called Form Factors which support different speeds and distances. Data center networks can be copper-based connections, fiber-based connections, or a combination of copper and fiber cables called hybrid connections, and transceivers can receive and transmit data in both copper and fiber optic cables.

Why Fiber over Copper

Fiber Copper
Made up of glass fiber Made up of copper wire
Carries data in the form of light Carries data in the form of electric signals
Offers higher bandwidth Offers lower bandwidth
Transmission speed is faster Transmission speed is slower
Low latency High latency
Installation cost is high Installation cost is less
Attenuation is very low Attenuation is high
Fiber is thin and light, easily breakable Copper is heavier and thick, difficult to break
More resistant to corrosive materials Prone to corrosive materials
More reliable, and durable Less reliable, and durable
Large life span Small life span
Not affected by electrical/magnetic interface Affected by electrical/magnetic interface
More secure – no leakage of light and are difficult to tap Less secure – leakage of signals and easy to tap
No Crosstalk problem problem Prevalent to Crosstalk problem
High noise immunity  Low noise immunity
Charge carriers are photons, which do not carry any charge, so they do not get affected Charge carriers are electrons, which carry a negative charge, so they get affected when they move in a wire

Standard Form Factors

The form factor states the physical dimensions of a transceiver which varies in size and shape depending on the speeds and protocols supported. Optical transceiver manufacturers design optics according to the Multisource Agreement (MSA). This is a standard for ensuring that the same form-factor transceivers from different vendors are compatible in size and function for interoperability with different vendor optics.

Type Speed
SFP 1G
SFP+ 10G
SFP28 25G
SFP56 50G
QSFP+ 40G
QSFP28 100G
QSFP56 200G
QSFP-DD 400G
OSFP/QSFP-DD 800G

SFP – Small Form-factor Pluggable

QSFP – Quad Small Form-factor Pluggable

QSFP-DD – Quad Small Form Factor Pluggable Double Density

OSFP – Octal Small Form-factor Pluggable

Transceiver Standards are based on the speed of the optics. Let’s take 100G transceivers that have different form factors CFP, CFP2, CFP4, CXP, and QSFP28. QSFP28 is the newest version of the 100G Optical transceiver. QSFP28 is a widely used 100G transceiver because of its high performance, lower power consumption, and higher density.

QSFP28

 Industry Standards Mode Distance Connector
QSFP28 Industry Standards Mode Distance Connector
100GBASE-SR4 IEEE 802.3bm, QSFP28 MSA, SFF-8665, SFF-8636, RoHS, CPRI, eCPRI MMF 100 m MTP/MPO-12
100GBASE-SR10 IEEE 802.3ba MMF 150m MTP/MPO-24
100GBASE-LR4 IEEE 802.3ba 100GBASE-LR4, IEEE 802.3bm, QSFP28 MSA, SFF-8665, SFF-8636 SMF 10km LC Duplex
100GBASE-ER4 IEEE 802.3ba, QSFP28 MSA Compliant SMF 40km MTP/MPO-12
100GBASE-ZR4 QSFP28 MSA Compliant SMF 80km LC Duplex
100GBASE-DR IEEE 802.3cd 100GBASE-DR Specification compliant SMF 500m LC Duplex
100GBASE-FR 100G Lambda MSA 100G-FR Specification compliant SMF 2km LC Duplex
100GBASE-LR 100G Lambda MSA 100G-LR Specification compliant SMF 10km LC Duplex
100GBASE-PSM4 100G PSM4, QSFP28 MSA Compliant SMF 500m MTP/MPO-12
100GBASE-CWDM4 IEEE 802.3ba, IEEE 802.3bm, SFF-8665, SFF-8636, 100G CWDM4 MSA, QSFP28 MSA SMF 2km LC Duplex
100GBASE-4WDM QSFP28 MSA Compliant SMF 10km LC Duplex
100GBASE-DWDM IEEE 802.3bm, QSFP28 MSA, SFF-8636, SFF-8024 SMF 80km CS Duplex
100GBASE-BiDi QSFP28 MSA Compliant SMF 20km LC Simplex
100G CLR4 100G CLR4 Industry Alliance SMF 2km CS Duplex

SR – Short Range

LR – Long Range

ER – Extended Range

ZR – Ze Best Range

LRM – Long Reach Multimode

PSM – Parallel Single Mode Fiber

WDM – Wavelength division multiplexing

CWDM – Coarse wavelength division multiplexing

DWDM – Dense wavelength division multiplexing

BiDi – Bidirectional optical transceiver

The Fanout

A high-speed port is broken into multiple low-speed ports are called Breakout Ports or Breakout cables. For example, a switch with a 400G port can be connected to 4x100G ports using breakout cables. Breakout cables are also called “fanout” cables.

Rate Technology Breakout Capable Electric Lanes Optical Lanes
10G SFP+ No 10G 10G
25G SFP28 No 25G 25G
40G QSFP+ Yes 4x 10G 4x10G, 2x20G
50G SFP56 No 50G 50G
100G QSFP28 Yes 4x 25G 100G, 4x25G, 2x50G
200G QSFP56 Yes 4x 50G 4x50G
2x100G QSFP28-DD Yes 2x (4x25G) 2x (4x25G)
400G QSFP56-DD Yes 8x 50G 4x 100G, 8x50G

Aviz ONES & The Optical Transceiver

Open Networking Enterprise Suite (ONES) is a network management and support application that offers the industry’s only multi-vendor, a multi-NOS solution that delivers Orchestration, Visibility, and Assurance – enabling SONiC adoption in new or existing deployments. ONES consumes telemetry from switches running SONIC, and other NOSs such as NVIDIA Cumulus Linux, Arista EOS, or Cisco NX-OS, and delivers deep insights into the link optics with information on

  • Inventory metrics for speed, type, breakout, lanes, manufacturer, etc.
  • Health metrics based on Digital Optical Monitoring (DOM) telemetry, an industry standard to access operating parameters of transceivers such as Tx Power, Rx Power, Temperature, Supply Voltage, Laser Bias Current, and much more.

This information is collected across any platform and the data is normalized to provide a unified view of the network fabric. ONES provides critical information on faulty transceivers in the fabric with drilled-down capabilities to identify the root cause of failures. An example of how deep ONES can go in terms of providing visibility for optics inventory and operational health is below.

Conclusion

Link availability is critical for any network operations, as the failure of a network link is the primary source of customer-impacting issues in the majority of cases. Hence, optics telemetry monitoring is a must-have in data center monitoring solutions to (a) understand link issues due to transient or permanent failures and (b) proactively identify the optics that are likely to fail and fix them ahead of time. Aviz ONES provides deep insights into optics to achieve the above goals. The optics data collected by ONES can also be extremely useful in qualifying optics before purchase decisions are made. Contact us to learn more about how Aviz can help with Optics Monitoring for your SONiC network.

Categories
All

Aviz Welcomes Zaid Kahn To Its Technical Advisory Board

Aviz Welcomes Zaid Kahn To Its Technical Advisory Board (1)

Aviz delivers the industry’s first multi-vendor stack for SONiC (Software for Open Networking in the Cloud) to help enterprises across the globe transition to the open-source NOS. We are incredibly grateful to have Zaid Kahn, who led the transition to SONiC at LinkedIn and now leads Cloud AI and Advanced Systems at Microsoft, join our team as a technical advisor. With Zaid joining our Technical Advisory Board, our customer experience empathy will increase and our expertise in cloud and AI will skyrocket, as we empower enterprises to leverage tools for multi-vendor SONiC deployment and operations.

“I am very honored to join Aviz technical advisory board. SONiC has proven to be the de-facto open source network operating system and Aviz is right there to enable faster adoption”

Zaid Kahn, GM, Cloud AI and Advanced Systems at Microsoft, and Board Member at OCP

Aviz is at the forefront of the SONiC enablement for the enterprise, who is looking to replicate what hyper-scalers have done, and benefit from the open-source NOS in terms of hardware choices, cost savings, and control they can exert on their network when compared to proprietary stacks. Aviz ONES, a key pillar of our multi-vendor SONiC stack, brings new capabilities for enterprises on top of the NOS for orchestration and visibility to utilize information and automation across their entire fabric.

“His experience of transforming a global enterprise network with SONiC and innovating in the Cloud and AI makes Aviz ecstatic to have Zaid on our Technical Advisory Board. Zaid’s insights and leadership will propel Aviz further in fulfilling our vision of enabling global enterprises to deploy and operate SONiC successfully and build out their Cloud and Network AI capabilities. Welcome, Zaid!”

Vishal Shukla, CEO, Aviz Networks

Aviz has built a global team of networking experts who have not only delivered products that are enabling SONiC, but also have made it more viable for using any ASIC or any switch underneath. Our goal is to usher enterprise into the data and AI driven new era of Networking 3.0 by leveraging SONiC as a platform to truly enable a multi-vendor networking cloud for the enterprise.

“Creating a Cloud and AI stack is no easy task in itself, but doing it across multiple ASICs, Switches, and NOSes brings a whole different challenge of normalizing data across all the underlying components. We have already made significant progress towards that objective, and Zaid’s vision with Open Cloud and AI based Networks shows us an even brighter future. I am honored for Zaid to be joining our team. It increases our confidence to build open networking Cloud and AI capabilities, which will be the advent of Networking 3.0. Welcome, Zaid!”

Chid Perumal, CTO, Aviz Networks

Aviz has been helping our lighthouse enterprise customers in several verticals with their SONiC deployment and operations for over 18 months now, both via our products and engineering services. The experience of helping customers successfully deploy SONiC on thousands of devices in their fabric is parallel to none, and it helps us to constantly evolve our products to catch the 5% issues that truly are business impacting. The fact that our products stream over 200 data points all the way from software to every single hardware component is no accident. It has helped us create the most comprehensive multi-vendor SONiC stack for the enterprise.

“Networks are evolving fast, and large scale data-driven networks are the future of networking. The biggest challenge we have is turning the ocean of data derived from them into actionable intelligence using AI. Our products, ONES and OPB, are not just providing a means to confidently transition to SONiC. They are evolving into the next generation tools that will be essential for operators to understand and manage their complex networks with ease. We feel overjoyed to have Zaid on our Advisory Board and to have his help to make our vision of Open-, Cloud-, and AI-First Networks a reality. Welcome, Zaid!”

Gautam Agrawal, CPO, Aviz Networks
Categories
All

Introducing ONES: The Supportability Stack for SONiC

We recently announced the general availability of Open Networking Enterprise Suite (ONES), the industry’s first supportability stack designed to empower network operators to migrate to SONiC. Since its inception, Hyperscalers have used open-source SONiC to manage and control their network. Enterprises globally are now looking to replicate the Hyperscaler success, but they face unique challenges around SONiC supportability as they transition to the open-source NOS. With Aviz’s ONES, and a growing multi-vendor SONiC ecosystem, enterprises can now easily transform their networks like hyperscalers.

Enabling SONiC Adoption for New or Existing Networks

SDXCENTRAL DEMOFRIDAY WEBINAR - WATCH NOW

ONES ushers network operators into a new era of open networking by allowing:

  • Hardware Agnostic Interoperability: ONES works on any Switch using any ASIC running SONiC, providing the software stack network operators need to make the move to white box switches in data centers and edge networks while delivering visibility for traditional NOSes such as Cumulus Linux, EOS, or NX-OS as well.
  • Deep Visibility and Control: Continuous uptime and system health begin with knowing your infrastructure and being in tune with your network fabric. ONES provides deep visibility down to the last detail on every component displaying dozens of metrics so you have the insights and power to drive peak performance while maintaining optimal system health.
  • 24/7 Multi-Vendor SONiC Support: Aviz utilizes the ONES application and a wealth of SONiC expertise to provide an immediate response to find and fix issues before they result in downtime. The Aviz Support for SONiC is backed by vendor SLAs, so enterprises can rely on a single entity for every platform deployed.
These new capabilities enable the enterprise to move swiftly in its migration to SONiC, making ONES the most comprehensive and inclusive solution that delivers end-to-end visibility for multi-vendor, multi-NOS networks.

    Hardware Agnostic Interoperability

    Most enterprises today are diversifying their hardware vendor portfolio in light of ongoing chip shortages or simply for optimizing their infrastructure cost. Quality of Service, user experience, and uptime across a diverse set of platforms are of grave concern when it comes to leveraging open-source software. We are proud to bring hardware-agnostic capabilities to SONiC deployment and operations. ONES delivers near real-time visibility across any SONiC (community version or distribution). Our telemetry agents stream and normalize data regardless of the underlying hardware and the version of SONiC it is running, and provide a unified view of the entire fabric.

    Figure 1: Dashboard of hardware/software components for every device in a multi-vendor network fabric

    Deep Visibility and Control

    No one wants to blindly adopt new technologies, especially open-source. ONES brings deep visibility with over 200 telemetry metrics collected via our agents in near-real time and provides dozens of operational and monitoring widgets that allow operators to gain insights into every aspect of their fabric, be it CPU/Memory utilization, SONiC microservices, or traffic errors. The biggest challenge in working with multi-vendor deployments is telemetry normalization before the data collected can be put to use for creating a unified view. We have worked with our early customers through proof of concepts and early deployment of ONES for almost a year now to bring the right level of visibility for large-scale multi-vendor SONiC operations.

    Figure 2: Topology view of every device and network connection in the fabric

    24/7 Multi-Vendor SONiC Support

    In our experience, the key issue for enterprises in adopting SONiC has been a lack of truly unified enterprise-grade support across various platforms. ONES not only brings the visibility, but also a wealth of SONiC expertise in our team that has been actively involved with the community for years. We have not just helped organizations deploy and test SONiC over the years, but also worked tirelessly to establish partnerships and SLAs with all major Switch and ASIC vendors to enable a unified channel of SONiC support. This is what completes ONES as the supportability stack for SONiC. The prowess of ONES is in the network effect of the benefits every enterprise gets from our collective experience of deploying SONiC for multiple use cases and resolving issues identified across multiple platforms.

    The SONiC Momentum Continues with ONES

    The General Availability of ONES, our current customers, our partners, ongoing deployments and pilots are not only a proof of the our momentum but also validates that we are changing the pace of SONiC adoption on white box switches in the enterprise. SONiC is no longer a buzz, it has quickly become one of the most sought-after technologies in the networking industry, and we are excited to contribute in creating value for the SONiC ecosystem. To learn more about ONES, watch Enabling SONiC Adoption for New or Existing Networks hosted by SDxCentral.

    Categories
    All

    Open-Source Alternative for the Network Packet Broker

    Server racks in computer network security server room data center. 3D render dark blue. Generative AI

    Network Packet Brokers (NPB) have been a critical component of the enterprise infrastructure for decades. They implement a range of tools to access, filter, and analyze traffic (a.k.a. network packets). NPB solutions collect packets from multiple links, filter them, and distribute them to tools specialized in performance monitoring, security analyses, and more by delivering packets specific to their needs. Typical NPB solutions are appliance-based, and proprietary in nature, making them expensive and inflexible. The cost of implementing such solutions in large data centers and edge networks can run into hundreds of thousands of dollars, if not more.

    The evolution of application-driven networking has been lauded by many industry leaders. Recent advancements in ASICs have increased their programmability, TCAM scale, and introduced flexible matches and actions. This creates opportunities to disaggregate the software and hardware components for packet broker solutions while meeting the performance and capacity requirements of the enterprise.

    Every data center in the world will deploy data processing units (DPU) to isolate the application layer from the control plane within five years. In every single data center, enterprise, or cloud, the control plane, and the application plane will be isolated. I promise you that.

    CEO, NVIDIA

    How can disaggregation be achieved for Network Packet Brokers?

    Before answering this question, we need to understand the core reason why NPB solutions are proprietary and appliance-based. Primarily it is the “unavailability” of a standardized NOS (Network Operating System), allowing vendors to utilize and exploit the power of their ASICs to deliver the capabilities required by packet broker solutions.

    With SONiC (Software for Open Networking in the Cloud), the open-source NOS, it is now possible to think of packet brokers as networking applications that harness the power of modern ASICs. SONiC has quickly become the standards-based open-source NOS that is being regarded as the “Linux of Networking” with support for over 100 SKUs across multiple ASICs. It provides a state-of-the-art microservices-based NOS architecture that paved the way for networking services as disaggregated applications on commodity hardware.

    What may a disaggregated Packet Broker solution look like?

    A network packet broker built using commodity switching hardware and open-source SONiC would be a truly disaggregated solution. NPB functions such as filtering, forwarding, and load balancing can be achieved using a programmable API, making such a solution truly software-defined. SDN (Software Defined Networking) principles can be used to program the ASIC with policies and services, such as filtering and traffic replication. The picture below provides a high-level depiction of a disaggregated packet broker.

    Figure1: Disaggregated Packet Broker

    What are the benefits of using a disaggregated Packet Broker solution?

    First off, a disaggregated packet broker solution based on open-source NOS eliminates the proprietary appliances significantly reducing the CapEx and OpEx. Secondly, it allows for the choice of hardware to leverage available speeds (from 10GbE to 400GbE) at a fraction of the cost of typical NPB solutions. Thirdly, the open-source nature of such solutions allows for easy integration with commercial analyzers available for performance and security monitoring. Last, but not least, disaggregation allows for the repurposing of hardware that is typically discarded during the network refresh cycles, making the solution even more lucrative, especially at a time when the semiconductor shortage has crippled the network infrastructure upgrade initiatives.

    The “Open Packet Broker” by Aviz

    ‍Aviz’s Open Packet Broker (OPB) is the industry’s first software-based containerized application built on top of the open-source SONiC to enable monitoring and security tools to access the network traffic. It is truly disaggregated because it can be deployed of your choice of Switch/ASIC hardware, as long as it supports SONiC. OPB enables you to easily scale up or down to meet the ever-changing needs of network visibility and security tools, empowering you to manage demand with maximum efficiency. To learn more about the Open Packet Broker schedule a demo.

    Categories
    All

    Semiconductor Shortage: Your Network Is Calling For A Multi-Vendor Strategy

    Closeup of electronic circuit board with CPU microchip electronic components background

    In the wake of the ongoing supply chain crisis, the one thing we have seen over and over is that availability of silicon is the single most important barrier for the enterprise trying to expand and/or upgrade their network infrastructure. Yes, every enterprise needs to transform digitally in this new world of remote-first work culture. Yes, there are complications with adopting new open-source architectures. And Yes, there is a shortage of talent globally. But that fact remains, that we can not even start our digital transformation initiatives if the semiconductor crisis continues to be where it is. Simply put, the enterprise needs to adopt a multi-vendor strategy so they can source hardware from where supply is.

    As the U.S. government noted in its “Briefing Room” blog, the paucity of semiconductors has not only been affecting the automotive industry, it has also been “dragging down the U.S. economy” and “could cut nearly a percentage point from GDP growth.

    HARVARD BUSINESS REVIEW, SCHUH, SCHNELLBÄCHER, TRIPLAT, AND WEISE

    ‍How can networking leaders of multi-million/billion dollar corporations deal with this global supply chain crisis that is plaguing their digital transformation initiatives?

    Relying on a single vendor who is constantly prioritizing and deprioritizing order fulfillment depending on how best they can juggle their customers isn’t cutting it now and will never be sufficient going forward. If Vendor A can only fulfill a portion of your hardware needs, the only option is to bring Vendor B, and Vendor C into the mix, from the very beginning, as part of your procurement strategy. But that brings up a different challenge.

    How would you standardize the software stack across multiple hardware vendors, when every vendor has traditionally provided its proprietary software stack?

    Fortunately, the “Open-Source” revolution in networking started long before the pandemic. We are now at a point where true disaggregation of hardware and software is possible with SONiC (Software for Open Networking in the Cloud). SONiC is an open-source network operating system based on Linux that runs on silicon from all major vendors. It is deployed today at multiple hyperscalers and web-scale enterprises for their Data Center and Edge Networks. The majority of ASIC/Switch vendors now support SONiC via their SAI/SDK integrations through standard APIs defined by a thriving community of SONiC developers.

    ‍Who would you go to, when your production network has issues that span across hardware from multiple vendors running “Open-Source” software that no one vendor truly owns?

    ‍It’s open-source. Community builds it, and the community maintains it. But, essentially the enterprise that is using it owns it and maintains it, which is fine as long as its core business is to sell networks (example: hyperscalers). When it comes to the non-hyperscalers, this is what scares them into going back to the proprietary stacks, and a single vendor strategy. And for this reason, network disaggregation cannot be limited to hardware and software segregation. We need new business models that can support “Open-Source” based stacks across multiple hardware platforms for the enterprise, much like how Red Hat supported enterprises through their adoption of Linux.

    ‍So, how would an entity like Red Hat support a multi-vendor strategy for network fabrics, without rolling out its own distro (much like RHEL)?

    This is where the new business model is needed. Customers crave quality, and quality doesn’t have to translate to a single SONiC distro across all hardware platforms. After all, each vendor spends tremendous amounts of R&D efforts in perfecting their own solutions. At Aviz, we recognize that customers care about their use cases. They are willing to embrace the nuances of the underlying hardware, and even software if there is a single point of accountability for 24×7 support.

    How is Aviz leveraging the “Open-Source” SONiC ecosystem and supporting this multi-vendor strategy?

    At Aviz, we are building “Open-, Cloud-, and AI-First Networks.” At the core of it is the support we offer for helping the enterprise not only through their multi-vendor SONiC deployments but also through post-production network monitoring, so you can be confident that there is someone you can call upon when you need help.

    We support multi-vendor SONiC deployments by:

    • Ascertaining quality: working with customers in identifying and documenting their use cases. We follow that by qualifying multiple vendors and certifying SONiC on their hardware in our lab environment for various speeds and feeds.
    • Providing 24×7 support: with our application-driven support model that leverages deep telemetry from devices running certified builds to proactively monitor your SONiC fabric, along with our team of SONiC experts that are available around the clock.
    • Enabling end-to-end SLAs: with both Switch and ASIC vendors to ensure that issues that trickle down to the platform drivers, or SAI/SDK layers, or the physical hardware itself for that matter can be addressed in a transparent and timely manner via well defined SLAs.
    • Delivering patches and upgrades: using our CI/CD-based sandbox environment to run regressions before the patches or upgrades are released for the customer’s production environment.

    Conclusion

    If there was ever a good time for the enterprise to consider a multi-vendor strategy for their network infrastructure, it is now. Rest assured that the entire ecosystem has evolved to a point where true disaggregation is not only feasible but the only viable option for catering to the new demands of digital infrastructure upgrades. Our customers are seeing immense value in our disaggregated application-driven support for “Open-Source” SONiC, which is a key driver to adopting a multi-vendor strategy and de-risking the challenges posed by the global supply chain crisis. To learn more about how our disaggregated application-driven support model works, schedule a discovery call with us.

    On a separate note, we are looking for dreamers like ourselves who can not only contribute code for multi-vendor SONiC development but also help expand our vision. Send us a note at hello@aviznetworks.com to start a conversation.

    Categories
    All

    Multi-Vendor SONiC: Creating New Opportunities For The Enterprise

    Repairman in server room between mainframes used for computationally intensive tasks such as scientific simulations or 3D rendering. Professional doing professional equipment inspection using tablet

    Software for Open Networking in the Cloud (SONiC, an open-source networking operating system) recently moved from the Open Compute Project (OCP) to The Linux Foundation (LF). This enables neutral governance, and further, accelerates its adoption in a multi-vendor-supported ecosystem.

    Leading open source network operating system enabling disaggregation for data centers now hosted by the Linux Foundation to enable neutral governance in a software ecosystem.

    THE LINUX FOUNDATION

    The LF provides the perfect venue for developer-led growth, and a public-private partnership across contributors, vendors, and users alike to nurture the open-source networking stack. It is clear that the future of Open-Source Networking is brighter than ever, given the traction SONiC has been seeing. Today, every large enterprise data center is considering SONiC deployment for the simplicity, agility, and reliability it brings above and beyond its cost advantage. More and more market segments such as edge, retail, campuses, and telcos are embracing SONiC at a supersonic speed. Their eyes are on the prospects of disaggregated networking and innovative applications in the cloud (hence software for open networking in the cloud).

    Let’s take a deeper look at the “Multi-Vendor” aspects of the Open-Source SONiC ecosystem, and how enterprises can benefit from it

    Typically, there are four different models when it comes to open-source software development:

    • Traditional Open-Source: when individuals or a group of developers work on interesting innovations with personal initiatives; and no involvement from public/private corporations. They typically just open-source their work and anyone in the community including public/private corporations can consume their work in whatever ways they like. Example: PANDA (Protocol And Network Datapath Acceleration).
    • Traditional Open-Source Distributors: when companies create businesses around traditional open-source projects and contribute some or all of their work back to the core of the project, but monetize some aspects of development and/or deployment. Example: Red Hat Enterprise Linux (RHEL).
    • Single-Vendor Open-Source: when a company aims to disrupt existing markets open sources its software to change the dynamics of the ecosystems around it and gain the mindshare of the community. Typically, in such cases, while the company builds a community around its technology, they also create an ecosystem for value-added products and services it can monetize. In doing so, they maintain strong control of and around their project and technology. Example: MongoDB.
    • Multi-Vendor Open-Source: when a community and multiple organizations come together with a common goal of building and standardizing a software stack across the entire ecosystem. In such cases, the collaboration amongst vendors and the community focuses on areas that are not directly a part of the participating organizations’ core assets. Example: Apache Kafka.

    The Future of Networking is Multi-Vendor Open Source

    SONiC, the open-source NOS, is one of the best examples of “Multi-Vendor Open-Source” software in networking, where multiple vendors, service providers, hyperscalers, and enterprises collaborate with the community and contribute to the future of Open-Source Networking. SONiC runs on Switches and ASICs from multiple vendors to offer a full suite of networking features via SAI/SDK layer integration that is specific to their core asset. Multi-Vendor Open-Source eliminates proprietary software dominance, giving users (enterprises) more of what they have been seeking for years: control and choice.

    Let’s do some math: Control + Choice + Zero Cost NOS + Community = immense value for the enterprise

    Add to this enterprise-grade support from Aviz, and you have everything you need to eliminate the proprietary software dominance from your network infrastructure.

    SONiC is the clear winner for which NOS to use across your network fabric and Aviz is the trusted partner in your SONiC journey because our goal is to make SONiC ubiquitous, intelligent, and accessible for every stakeholder. Aviz is the moderator of the SONiC Community, the co-creator of the SONiC CAB (Customer Advisory Board), and we don’t sell ASICs, Switches, or SONiC Distros. Neutral, impartial, and packed with SONiC Experts, Aviz is here to help you implement, manage, and improve your Open Networking infrastructure.

    Here are 3 ways to maximize your benefits from the SONiC Community and Aviz:

    1. Engage in the SONiC Experts group on LinkedIn
    2. Steer development in a direction that creates enterprise value in the SONiC CAB: sonic-cab@aviznetworks.com
    3. Contact Aviz to derive the greatest benefits from SONiC deployment

    SONiC has the potential to transform your network infrastructure much like Linux transformed Compute. Working with Aviz exponentially grows the value you can derive from the SONiC transition. Schedule a discovery session with me to learn more about how Aviz can help.

    Categories
    All

    Open vs Open-source: A Paradigm Shift in Networking

    The networking industry saw a huge number of startups evolve in the last decade. Unfortunately, the vast majority of them did not deliver on the promise of creating a truly open and disaggregated networking ecosystem, which is evident from the consolidation that followed their evolution. Fortunately, two broad concepts in networking were clear winners: 1) Software Defined Networking (SDN), and 2) Disaggregation in Networking solutions. SDN started off well with controller (NSX) and non-controller (EVPN) based methods and is now used as the basis of orchestrating VPC (Virtual Private Cloud) networks in public clouds. Disaggregated Networking as a concept was popularized by companies such as Cumulus, which helped immensely in generating the demand for standards-based solutions across multiple ASIC and Switch vendors. What did not change, however, was the networking protocols, and to put it simply – BGP remains BGP, even after decades of innovations and transformations.

    In 2017, Microsoft and the Open Compute Project (OCP) rolled out SONiC (Software for Open Networking in the Cloud) as a free and open-source network operating system based on Linux. It allows cloud operators (public, and private) to share the same software stack across hardware from different vendors. As SONiC started gaining popularity, the disaggregated solutions started referring to themselves as Open Networking solution providers. The word ‘Open’ makes sense in the context that customers are open to using any hardware, but it also creates a false pretense that the whole solution is open-source, which was never the case. As someone who used to sell disaggregated networking solutions from some of the aforementioned vendors, I was often asked by customers to provide code-level access, since the solutions were advertised as ‘Open Networking’, only to find out that  ‘Open Networking’ did not really mean  ‘Open-Source’.

    As true believers of the ‘Open Source Networking’ paradigm, we not only contribute towards SONiC’s development but are honored to be managing the SONiC Development Program as a neutral entity in the SONiC ecosystem.

    VISHAL SHUKLA

    Fast forward to today, with the level of maturity and adoption SONiC has achieved, we (at Aviz Networks) believe that now is the right time for the industry to start delivering on the promise of truly ‘Open Networking’ a.k.a ‘Open-Source Networking’. If you are looking to transition to truly Open Networking, our recommendation is to consider the following when evaluating options:

    ‍Is the ‘Open Networking’ stack forcing a vendor lock-in?

    In the early 2000s, networking solutions used to be completely proprietary, primarily from the traditional networking companies. In the next decade, disaggregation happened, and a choice of the white box was made available, which was a welcoming step forward for the industry. But there was still a lock on the NOS, which was not only closed source but also came with a hefty price tag. The huge licensing fee for commodity features will continue until we move towards an Open-Source NOS as an industry, much like how Compute and Storage adopted Linux. ‘Open Networking’ solutions must work as a truly open stack – from the ASIC to the NOS. If there is any proprietary component in any of the layers, specifically on the software side, the whole solution locks you into a vendor-specific ecosystem.

    Is the ‘Open Networking’ solution future-proof?

    In the past decade – a series of proprietary NOS acquisitions by Switch/ASIC vendors created a deadlock for customers, which clearly shows that any solution sold under the conventional tag of ‘Open Networking’ is not future-proof. Customers who deployed proprietary NOS  on non-acquirer/competing vendors’ ASICs were left with no choice. They essentially have to either go back to traditional networking companies or buy another expensive NOS distribution. We see some NOS vendors committing to avoid such deadlocks, but they are also using the same proprietary NOS approach. Simply put, the cycle is repeating, and as long as there is proprietariness in the NOS layer, the networking stack will never be future-proof.

    ‍Is the ‘Open Networking’ solution only about Switch & NOS? What about Cloud and Apps?

    Today, a vast majority of providers (including public cloud), talk about disaggregated applications that can run on any infrastructure as long as there is a standard OS layer. This is a proven model within the ‘Compute’ and ‘Storage’ ecosystem (with Linux) and for the Mobile platforms (with iOS and Android). We recommend that the NOS you select should adhere to standards adopted vastly by the majority (if not all) vendors, and has the ability to drive innovation for networking applications, and cloud adoption.

    Shifting to Open-Source

    Truly Open Networking solutions must work as an ‘Open Source’ stack – from the ASIC to the NOS and Truly Open Networking solutions must work as an ‘Open Source’ stack – from the ASIC to the NOS and be standards-based for nurturing application and cloud-focused innovation in the network infrastructure. Today, we see SONiC as the only truly Open Source NOS that allows you to build your own networking stack, specific to your needs, and gives you the ultimate flexibility of using any ASIC or any Switch, all without locking into a vendor-specific ecosystem. Additionally, since SONiC makes the necessary networking protocols available as open-source software, it forces the license-based consumption model to move towards a pay-for-support and pay-for-apps ecosystem, a mode that has been widely adopted in Compute and Storage already. This is what ‘Open Networking’ promised, but Open-Source SONiC is delivering. Once implemented in your Data Centers, Campuses, and Edge Networks, SONiC will not only give you the control you always wanted, but it will also result in huge TCO savings that can be reinvested into your growth.

    At Aviz Networks, we are committed to enabling our customers and our partner ecosystem with ‘Open Source Networking’ via SONiC. As true believers of the ‘Open Source Networking’ paradigm, we not only contribute towards SONiC’s development but are honored to be managing the ‘SONiC Development Program’ as a neutral entity. We are working with several customers to make their SONiC deployments successful by offering SONiC enablement products and services that address critical gaps in SONiC adoption at the moment. We even provide containerized microservices on top of SONiC that are first of their kind networking applications to turn your hardware-based network functions (e.g. Network Packet Broker Taps and Aggregators) into software-only modules. Connect with us to learn more about our products and services or if you are interested in a demo.

    Open vs Open-source: A Paradigm Shift in Networking

    The networking industry saw a huge number of startups evolve in the last decade. Unfortunately, the vast majority of them did not deliver on the promise of creating a truly open and disaggregated networking ecosystem, which is evident from the consolidation that followed their evolution. Fortunately, two broad concepts in networking were clear winners: 1) […]