Community
Community

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

Category: Fabric Test Automation Suite

Your blog category

Categories
Fabric Test Automation Suite

The Missing Layer in AI Infrastructure: Why AI Networking Must Be Full-Stack, Open, and AI-Operated

AI infrastructure is growing at an unprecedented pace. Enterprises are racing to build clusters of GPUs, scale up AI workloads, and modernize their data pipelines. Yet one critical layer is often overlooked in these initiatives: the network.

While AI and data leaders focus on compute, storage, and models, the network quietly becomes the bottleneck. Traditional, static networks—built for legacy application traffic—can’t handle the dynamic, latency-sensitive, high-throughput demands of distributed AI workloads. And without visibility, orchestration, and automation across the full stack, enterprise IT leaders are flying blind in one of the most critical infrastructure domains of the decade.

At ONUG, where the community has long championed open, cloud-scale networking, this challenge is both familiar and urgent. It’s time to reframe the conversation: AI networking isn’t a peripheral concern—it’s the missing layer in AI infrastructure. And the solution is not just faster switches. It’s a full-stack, open, and AI-operated networking layer designed for the AI era.

The Blind Spot in AI Buildouts

Enterprise AI infrastructure is hitting production, but many organizations are discovering that their network isn’t ready.
Why? Because today’s networks were never designed for high-volume traffic, dynamic scaling of AI, or the precision tuning required for GPU interconnects. Most enterprise networks lack native support for lossless transport, multi-tenancy, or real-time visibility—all of which are essential when running distributed AI workloads where every millisecond counts.
Meanwhile, proprietary stacks slow down innovation. Observability is fragmented. Upgrades are risky. Operators are juggling CLI scripts, YAML files, and ticketing systems just to troubleshoot basic issues.
This isn’t sustainable. And it’s not how AI infrastructure should operate.

A New Layer: Full-Stack AI Networking

To solve this, enterprises need to rethink how networks are built and managed—starting with a full-stack approach.

We break this down into two key areas:
  • Networks for AI: The physical and virtual infrastructure optimized to connect GPUs with high-throughput, low-latency, lossless configurations.
  • AI for Networks: Intelligent automation powered by AI that simplifies Day 0–2 operations, from deployment to troubleshooting to compliance.
This full-stack model includes:
  • Open Network Operating Systems (NOS) like SONiC and Cumulus that decouple software from hardware
  • Multi-vendor orchestration layers that unify fabrics across OEMs
  • Observability and telemetry frameworks—offering deep packet inspection, metadata extraction, and visibility across 4G/5G/AI fabrics
  • LLM-based copilots that assist with upgrades, audits, performance tuning, and real-time issue resolution
Whether you’re deploying a reference architecture like NVIDIA Spectrum-X or an open fabric with SONiC, this approach ensures you’re not building AI infrastructure on a 20th-century network foundation.

Why Open Matters More Than Ever

Vendor-neutrality isn’t just a cost issue—it’s a control issue. The more proprietary your stack, the slower you move.
Open platforms like SONiC enable IT teams to:
  • Tune the network stack based on their specific AI workloads
  • Replace and upgrade hardware without rewriting the orchestration playbook
  • Integrate seamlessly with observability tools, automation platforms, and security frameworks
We recently hosted a PlugFest that brought together leading switch vendors, solution providers, and enterprise users to test and validate SONiC-based fabrics. The takeaway? Open networking is no longer an experiment—it’s ready for enterprise AI at scale, and it’s being certified and hardened by the community.

From Complexity to Clarity: AI-Powered Operations

Operating AI infrastructure shouldn’t require navigating dozens of tools or relying on tribal knowledge. Networks must evolve to support simplified, AI-powered operations.

That means:
  • Unifying management across the operations
  • Leveraging real-time telemetry for proactive troubleshooting
  • Automating repetitive tasks like compliance checks, and performance audits
  • Using copilots to generate insights, summaries, and reports that accelerate time to resolution
This is the future of AI networking—simplified, scalable, and guided by data.

Build the Right Layer

The network is where performance, cost, and reliability intersect—and where you can gain or lose the most.

The time is now to invest in AI networking as a full-stack discipline—not a siloed afterthought. By embracing open, AI-powered, and multi-vendor infrastructure, IT leaders can finally align the network with the speed of innovation in AI.

Meet us at ONUG Dallas, May 28–29. Stay ahead of the curve—book a 1:1 with our experts and see how Aviz accelerates AI networking
Book a meeting
Categories
Fabric Test Automation Suite

Simplify Your SONiC RMA Experience with ONES Backup & Restore

In today’s fast-paced networking landscape, data is a critical asset. Unexpected failures can lead to downtime, operational disruptions, and misconfigurations. When a network device crashes, engineers need a reliable backup to restore it quickly. Without structured backup and restore mechanisms, organizations risk prolonged outages and inefficiencies. This overview underscores the importance of regular backups and explains how ONES Fabric Manager Backup & Restore streamlines the process, ensuring seamless recovery in multi-vendor environments.

The Importance of Backup & Restore in Network Resilience

Backup and restore processes ensure rapid recovery from failures by preserving critical network configurations. Key components include:
  • Configuration Snapshots: Capture and store complete network configurations, including interfaces, QoS, ACLs, and VxLAN, enabling full restoration when needed.
  • Multi-vendor Compatibility: Ensures seamless backup and restoration across different vendor devices.
In RMA scenarios, replacing faulty hardware is only the first step—the real challenge lies in restoring the original configurations. Without a recent backup, administrators must manually reconfigure the failed switch, resulting in extended downtime, increased risk of errors, operational disruptions, and higher recovery costs due to additional troubleshooting and resource allocation.

ONES Backup & Restore: The Lifeline for Uninterrupted Networks

ONES Fabric Manager Backup & Restore ensures seamless recovery by securely storing configurations in a persistent Docker volume, enabling quick restoration, and eliminating manual reconfiguration. With pre-replacement snapshots for ZTP or upgrades, it offers a reliable rollback option. Designed for multi-vendor compatibility, it minimizes downtime, reduces risks, and streamlines RMA processes for efficient, error-free network management.
Figure 1: Backup Taken After Configuration

Streamlined Backup & Recovery Process

ONES Fabric Manager Backup & Restore captures essential configuration files (config_db.json, frr.conf, fmcli_db.cfg) by enabling both manual and automatic snapshot creation during key operations like reboot, ZTP, or image upgrades. Each snapshot is tagged with a timestamp or custom label for easy identification and restoration. In the event of a failure, users can quickly revert to a known-good configuration—minimizing downtime and eliminating the need for complex manual recovery steps.
Figure 2: Backup Management Page
Figure 3: Restore Management Page

Multi-Vendor Support for Diverse Environments

Designed for flexibility, ONES Fabric Manager Backup & Restore works seamlessly across various network devices. Its consistent and reliable backup and recovery capabilities make it an ideal solution for dynamic, multi-vendor infrastructures, ensuring uninterrupted network performance regardless of vendor diversity.
Book a demo today — because every second of network downtime costs more than you think.
Book now
Categories
Fabric Test Automation Suite

Introducing Aviz Service Nodes R2.0 – Empowering Enterprise, Data Center and Telco Deployments

In today’s ever-evolving network landscape, staying ahead means continuously innovating. Building on the trusted foundation of our previous releases—including the robust telco features of 1.3.1—we are excited to introduce R2.0. This release is not just an update; it’s a reimagining of how enterprise and telco networks operate, bringing you enhanced performance, precision, and peace of mind

What’s New in R2.0

In R2.0, we’ve focused on adding tangible value for our customers. Here’s a quick look at what you can expect:
  • Enterprise & Telco Readiness: Fully ASN-ready for seamless deployment across large-scale networks.
  • Packet Deduplication for DC: Optimizes network performance by eliminating redundant packet processing.
  • Advanced DPI for IP Packets: Leverages cutting-edge deep packet inspection for precise application identification.
  • Enhanced Packet Capture: Delivers robust support tailored for both telco and enterprise environments.
The image below provides a concise overview of ASN.

5G Core Advancements

ASN 2.0 is designed to handle the scale and complexity of both enterprise and telecom networks, ensuring seamless deployment and operational efficiency. It processes incoming telecommunication traffic, extracting critical metadata from 4G-LTE, 5G-NSA, and 5G-SA networks, providing end-to-end visibility for mobile operators and service providers.
With intelligent control-plane (C-Plane) and user-plane (U-Plane) correlation, ASN enables precise tracking of subscriber sessions, mobility events, and service quality metrics. Enhanced KPI monitoring now delivers sub-5-second precision for bandwidth and latency calculations, ensuring real-time network optimization. Additionally, optimized Kafka metadata export improves data accuracy and reduces processing latency, enabling faster insights and seamless analytics across modern network infrastructures.

Packet Deduplication

Our new packet deduplication feature is designed to streamline data flow. Removing redundant packets not only boosts processing efficiency but also improves real-time network analytics, ensuring your systems run at peak performance.
  • Deduplication Process: Traffic is continuously analyzed in real-time with a granularity of up to 8 milliseconds, enabling the rapid detection and elimination of duplicate packets. The analysis process begins by inspecting ingress packets for familiarity based on their type and fields. If a packet is found to exist within the customer-defined time frame, it is identified as a duplicate and dropped. Users can configure the Packet Source (Full or Routed Packet) to filter out redundant traffic, even across multiple hops. Anchor & Offset settings define where and how packet data is compared, while the Window Size (2-8ms) ensures efficient real-time deduplication. For detailed configuration, refer to the GitBook.
Below is the FlowVision UI snapshot for Packet Deduplication

Application Identification with Advanced DPI

With ASN 2.0, we extend beyond telco packet correlation to deliver end-to-end IP network observability, making ASN a powerful solution for data centers, FTTH, edge, and campus networks. By leveraging Deep Packet Inspection (DPI), ASN can now classify over 500 applications at line rate, extracting critical network intelligence for enhanced traffic analysis and security. ASN’s advanced DPI engine uses pattern recognition, SNI matching, IP geolocation, and port-based analysis to identify applications with high accuracy. It also captures user agent metadata, bandwidth usage, and key performance indicators (latency, packet loss, and jitter), enabling operators to optimize network performance in real time.

Additionally, enhanced KPI monitoring in ASN 2.0 delivers sub-5-second precision for bandwidth and latency metrics, while optimized Kafka metadata export ensures faster processing and improved analytics. Here is a snapshot of our Kibana dashboard, showcasing metadata extracted by ASN along with application insights identified by ASN’s advanced DPI engine.

Comprehensive Packet Capture

Effective troubleshooting and network analysis are at your fingertips with our enhanced packet capture capabilities. Designed to meet the rigorous demands of both enterprise and telco deployments, this feature empowers your team to quickly diagnose and resolve issues. ASN 2.0 allows users to configure interface capture settings, selecting specific ports for Tx/Rx traffic analysis. Capture duration can be set between 1-10,000 ms, with a file size limit of 1-1,000 MB. Users can define filenames for captured data and enable automatic cleanup of old PCAP files. For detailed configuration, refer to the GitBook.

Enhanced packet capture capabilities in ASN 2.0 enable users to generate live traffic in PCAP format for real-time network analysis. Designed for both enterprise and telco environments, this feature streamlines troubleshooting by facilitating faster diagnostics and issue resolution

Below is the FlowVision UI snapshot for Packet Capture

Real-World Impact: Benefits You Can See

Imagine a network where every packet is optimized for performance, every application is clearly identified, and troubleshooting is more efficient than ever. R2.0 is designed to empower your team, reduce operational costs, and provide the reliable network performance that modern enterprises and telcos demand.

R2.0 isn’t just an upgrade—it’s our commitment to continually pushing the boundaries of network observability. We’re excited for you to experience these innovations firsthand and look forward to supporting your journey into the future of enterprise and telco networking.

Book your demo to see how ASN R2.0 transforms network observability, optimizes traffic, and enhances real-time analytics for enterprise and telco deployments!

Book Your Demo
Categories
Fabric Test Automation Suite

Cutting Through the Noise: Packet Deduplication in Data Center Monitoring

Abstract

In modern data centers, network observability plays a crucial role in performance monitoring, security, and analytics. However, redundant network traffic can create inefficiencies, leading to excessive processing overhead and degraded system performance. This blog discusses an innovative approach to Network Observability DeDuplication, leveraging filtering, deduplication, and optimized distribution strategies to enhance network efficiency and improve analytics accuracy.

Introduction

As enterprises scale their digital infrastructure, managing high-bandwidth network traffic efficiently is a significant challenge. Duplicate packets from various sources, including TAPs and mirrored configurations (SPAN, ERSPAN), lead to unnecessary data processing, storage, and bandwidth consumption. Network Observability DeDuplication addresses this challenge by filtering irrelevant traffic, identifying duplicate packets, and ensuring optimized data distribution for analytics tools.

Solution Architecture

The proposed Network Observability DeDuplication solution consists of four key components:

Data Sources

  • Data Center Fabric: Network traffic is captured and forwarded through physical TAPs and mirror configurations such as SPAN and ERSPAN.
  • This traffic contains both relevant and redundant packets that need to be processed efficiently.

Filtering Fabric

  • Open Packet Broker Network Operating System (OPBNOS) ensures that only relevant traffic is forwarded, reducing unnecessary processing and enhancing system performance.
  • This step helps in reducing computational and bandwidth costs associated with duplicate or irrelevant data.

Core Fabric & Deduplication

  • Core Fabric: The filtered traffic is directed towards a high-capacity core fabric (100G-400G) for further processing.
  • Aviz Service Node (ASN): A specialized x86 + DPDK-based service node is utilized for high-speed deduplication.
  • Deduplication Process: Traffic is continuously analyzed in real time with a granularity of up to 8 milliseconds, enabling the rapid detection and elimination of duplicate packets. The analysis process begins by inspecting ingress packets for familiarity based on their type and fields. If a packet is found to exist within the customer-defined time frame, it is identified as a duplicate and dropped.

    Once the deduplication is complete, the refined traffic is seamlessly reintegrated into the core fabric for further processing and routing. By ensuring that only unique packets are forwarded, this approach enhances resource allocation, reduces processing overhead, and maintains low latency. The result is a streamlined traffic flow that improves data transmission efficiency, enhances application performance, and ensures a smoother user experience across the network.
  • Configuration Elements:
  • Packet Source: This defines what data fields to take in while analysing the packet for duplicates. Currently there are two packet source types:
  • Full Packet: The packet bytes from start to end, maximum of offset bytes.
  • Routed packet: In real time, the packets can be routed across multiple devices/hops. Hence there are chances the duplicate packets can be received to monitoring fabric with different Src MAC/TTL/Checksum fields. This option helps in finding duplicate packets irrespective of Routing parameters.
  • Anchor: This tells where to start reading the packet from. The possible values are Packet start ,L3 Start, L4 Start and L4 Payload.
  • Offset: To provision how many bytes are considered for duplicate comparison. Allowed Offset is 14 to 128 Bytes.
  • Window Size: This config tells the core logic to look for familiar packets in the 2,4,6 and 8ms range currently. If a user for example sets the window size to 6ms, the adjacent packets falling in each 6ms window will be checked for duplicate data.

Distribution Fabric

  • The deduplicated traffic is load-balanced and distributed across multiple analytics tools for network monitoring, security analysis, and performance optimization.
  • This ensures efficient utilization of analytics resources without redundant data inflating storage or processing costs.

Key Benefits

Performance Optimization

  • Reduces the overhead of processing duplicate packets, improving monitoring efficiency by 30 to 50%.
  • Enhances the responsiveness of analytics tools by providing cleaner, non-redundant data.

Cost Savings

  • Minimizes bandwidth and storage costs by filtering and deduplicating redundant traffic.
  • Optimizes computational resources by reducing unnecessary packet processing.

Improved Security & Compliance

  • Enhances security monitoring by providing accurate traffic visibility.
  • Facilitates compliance with regulatory requirements by ensuring only relevant network data is retained.

Conclusion

Network Observability DeDuplication is a transformative approach to handling large-scale network traffic efficiently. By integrating filtering, deduplication, and intelligent distribution mechanisms, organizations can achieve optimized network monitoring, reduced infrastructure costs, and enhanced analytics accuracy. As data centers continue to evolve, deploying such intelligent traffic management solutions will be key to maintaining high-performance, scalable, and secure network infrastructures.

Optimize your network monitoring with intelligent deduplication—because efficiency matters. Book a demo today!
Book Your Demo
Categories
Fabric Test Automation Suite

ONES AI-Assistant for Conversational Troubleshooting

We are excited to announce the launch of the AI Assistant feature in ONES 3.1, marking a major step forward in simplifying and enhancing network management. This new feature provides users with an intelligent conversational interface to interact with network health and inventory data. The AI Assistant can respond to a wide range of queries, streamlining the management process and offering critical insights instantly. The AI Assistant runs on-premises and functions efficiently on a CPU, without requiring a GPU or any tokens to process user queries.

AI Assistant Capabilities and Features

The AI Assistant is designed to handle technical queries related to network health and inventory management. Below are its core capabilities:

  • Network Health Insights: The AI Assistant can instantly provide you with information on network performance, including device status, uptime, and any potential anomalies.
  • Inventory Management: Whether you need to check on device configurations, availability, or health, the Assistant can offer real-time answers to help you manage your inventory efficiently.
  • Markdown and Text Responses: Answers are delivered in Markdown tables for organized data presentation, and regular text for straightforward inquiries, ensuring clarity and readability.
  • Stop Streaming Responses: If you prefer not to wait for the entire response to load or if the information provided is enough, you have the ability to stop streaming and receive a final answer instantly.
  • Recommended Questions: After each response, the AI Assistant will suggest related queries to help guide your next steps, ensuring you have all the information you need to make informed decisions.

How It Works

The AI Assistant is built on Generative AI and powered by advanced language models, which enables it to process complex network-related questions and generate responses in real-time with high accuracy and natural language fluency.

AI Assistant in Action: Real-Life Applications

The AI Assistant is designed to simplify day-to-day tasks and make data-driven decisions more accessible. Here are a few practical examples:
  • Real-Time Network Monitoring: Ask the Assistant, "What is the network summary for the last 24 hours" and it will immediately provide a response detailing device utilizations, packet drops if any present etc.
  • Inventory Management: Inquiring about the fan fault information? The Assistant can respond instantly with details like, “Fan details,” including when it stopped and its current condition, whether it’s working or not.

Conclusion

The AI Assistant in ONES 3.1 is poised to transform how users interact with their network and inventory data. Powered by Generative AI and the advanced language model, this feature offers a new level of convenience, efficiency, and intelligence to your network management tasks.

Book your demo to experience AI-driven network troubleshooting—gain instant insights, streamline inventory management, and simplify operations effortlessly!
Book your Demo
Categories
Fabric Test Automation Suite

Love the Idea of Multi-Vendor, But Don’t Want the Risk? Start with Network Observability

Let’s be honest — every network leader today has asked the question:

“Can I go multi-vendor?”

And just as quickly, the doubts follow:
  • What if support becomes a nightmare?
  • What if performance drops?
  • What if my team can’t manage the complexity?
  • What if something breaks?
Fair concerns. After all, the network is mission-critical.
But here’s the answer:

Start where the risk is low, impact is high, and the insights are immediate. Start with your network observability stack.

Why Visibility Is the Ideal Place to Start Your Multi-Vendor Journey

1. It's Passive by Nature.

You’re not routing packets. You’re observing them. That means no disruption to traffic, no downtime risk, and no changes to your core infrastructure.
You get full visibility into what’s happening across your network — without putting it in harm’s way.

This is what we call a calculated risk. The kind smart network leaders take.

2. You Get to See Multi-Vendor in Action — Before Going All-In

Standardize your packet broker and service nodes at the visibility layer. Use software-first, vendor-neutral platforms like Aviz that can plug into any environment — Cisco, Arista, NVIDIA, Broadcom, and more.

This gives you a real-world look at:
  • How multivendor really behaves
  • How support flows across vendors
  • How your ops team handles integration
  • How much you actually save

Spoiler: it’s often 50%+ in direct costs and 70%+ ROI over the lifecycle of the solution.

3. Your Data Center Footprint Shrinks

When you replace FPGA-based appliances with software-powered visibility nodes, running on commodity hardware or CPUs (and accelerating only when needed using DPUs like NVIDIA BlueField), the results speak for themselves:

  • Less space
  • Less power
  • Less cooling
  • Fewer boxes

One of our large telco deployments — in the capital region of one of the world’s biggest economies — saw an 80% reduction in hardware footprint, while supporting 30M+ subscribers with data granularity every 5 seconds.

That’s not an upgrade. That’s a transformation.

And Even If It Doesn’t Work — Your Core Network Is Still Untouched

That’s the beauty of starting with visibility.
Worst case? You gain insight, experience, and metrics — and you go back to your old model. Your core routing, switching, and orchestration aren’t affected. Your customers never notice. Your ops team just got smarter.
Best case?
  • You prove out the multi-vendor model in a real-world setting.
  • You gain flexibility.
  • You break vendor lock-in.
  • You become future-ready.

This Is How Every Disruption Begins

Most OEMs didn’t enter the network through the front door. They started where the stakes were lower — passive networks, test environments, monitoring layers.

Visibility was the Trojan Horse.

Now it can be your bridge to innovation.

If You’ve Ever Thought About Going Multi-Vendor — This Is Your Moment

Try it. Measure it. Stress-test it.
And then decide from a position of data, not fear.

Start with visibility. Make it software-first. Standardize the layers that matter.

Let Aviz show you what’s possible — safely, intelligently, and cost-effectively.
And Here’s How We’re Already Doing It

A leading telco, serving over 30 million subscribers in the capital region of one of the world’s largest economies, partnered with Aviz to modernize their network observability stack.

They took the calculated risk.

They replaced expensive, proprietary hardware with a multi-vendor, software-defined visibility platform — powered by Aviz.

The result?

  • 80% reduction in hardware footprint — saving on capex, power, and data center space
  • 50% savings in operational expense — driven by standardization and automation
  • 5-second telemetry granularity — at scale, across vendors
  • Zero disruption to their core network
They now have the proof — multi-vendor visibility works. It’s efficient. It’s scalable. And it opens the door to future innovation across the entire network.

This is how you start.

Low risk. High reward.
The first step to breaking free from vendor lock-in and building an intelligent, open, AI-connected network.

Let’s talk. Let’s try. Let’s build the future — one layer at a time.

Categories
Fabric Test Automation Suite

Cisco 8000 + SONiC with Aviz ONES Bootcamp – Why You Should Join?

Why You Should Join?

Ready to unlock the future of networking? If you’re managing AI workloads, optimizing data center operations, or just curious about the power of SONiC on Cisco 8000, then this is the event you don’t want to miss.

AI is changing everything – from high-performance computing to real-time analytics, and your network needs to keep up. That’s where Cisco and Aviz Networks come in. Together, we’re redefining AI-ready infrastructure with SONiC-powered networking – delivering agility, scalability, observability, and orchestration.

Join us for the Cisco 8000 + Aviz ONES SONiC Bootcamp on April 3, 2025, where we’ll break down how to run AI-optimized networks with SONiC on Cisco hardware.

Why Should You Care?

Imagine a network that’s not just fast, but intelligent – make your network self-optimized for AI workloads, and that gives you complete control over traffic flows. That’s the power of SONiC on Cisco 8000, supercharged by Aviz ONES.
If you’re a network architect, engineer, or AI infrastructure leader, this bootcamp is designed for you. Whether you’re considering SONiC for the first time or already deploying it, we’ll show you how to make it work seamlessly on Cisco 8000.

What’s on the Agenda?

Part 1: The Cisco 8000 SONiC Evolution

(Presented by Cisco)

  • Why SONiC on Cisco 8000? The competitive advantage of open networking on Cisco’s high-performance hardware.
  • Deployment Models: Learn the best ways to integrate SONiC into your AI-driven infrastructure.
  • Real-World Success Stories: Discover how organizations are using SONiC on Cisco 8000 to scale their networks.

Part 2: AI-Driven Operations & Observability with Aviz ONES

(Presented by Aviz)

  • Live Demo: See how Aviz ONES provides deep observability and orchestration for Cisco 8000 SONiC environments.
  • Automated Deployment & Provisioning: Speed up your SONiC deployment with seamless automation.
  • Real-Time Monitoring & Troubleshooting: How AI-driven telemetry helps you detect and fix issues before they impact performance.

AI Networking with SONiC: A Practical Guide

  • Optimizing SONiC for AI workloads – tuning for congestion control, RoCEv2, and high-bandwidth applications.
  • Scaling AI clusters on Cisco 8000 SONiC fabrics – designing high-efficiency networks.
  • Implementing PFC & ECN for network-wide traffic control – minimizing latency and maximizing throughput.

Join Us – Register Now!

  • Date: April 3, 2025
  • Format: Online
Register Now
Categories
Fabric Test Automation Suite

5 Myths About SONiC NOS

The rise of SONiC (Software for Open Networking in the Cloud) has disrupted the traditional networking industry, offering enterprises a vendor-agnostic, open-source alternative to proprietary NOS solutions. But with disruption comes misconceptions, and SONiC is no exception.

Let’s debunk the top five myths about SONiC NOS and set the record straight.

Myth #1: SONiC is Only for Hyperscalers

Reality: SONiC is Ready for Enterprises, AI Workloads, and Beyond

While hyperscalers like Microsoft pioneered SONiC, it’s no longer just for the tech giants. Enterprises across healthcare, retail, finance, and AI-driven data centers are deploying SONiC to cut costs, increase flexibility, and escape vendor lock-in.

👉 2025 PlugFest validated SONiC for enterprise-grade use cases, including Layer 2 networking, AI fabric, and PoE-enabled whitebox switches—proving its viability beyond hyperscalers.

Want to hear real-world enterprise SONiC success stories?

Join us for a compelling SDxCentral panel discussion featuring insights from our customers, Techevolution and 1984, alongside our partner, EPS Global. Explore how SONiC is transforming enterprise networking with unparalleled flexibility and efficiency.

Myth #2: SONiC is Difficult to Deploy

Reality: SONiC Adoption is Faster and Easier Than Ever

There’s a misconception that SONiC requires deep coding skills or hyperscaler-level engineering resources. That was true in its early days, but today, we have 1-click SONiC fabric deployment and  enterprise-ready SONiC solutions come with automation, APIs, and user-friendly management tools.

💡 One-click SONiC migration guides and vendor support services have made the transition seamless—offering enterprises the same ease of deployment as traditional NOS solutions.

Myth #3: SONiC Lacks Vendor Support

Reality: A Thriving Ecosystem Backs SONiC

Some believe that using an open-source NOS means you’re on your own. That couldn’t be further from the truth. The SONiC ecosystem includes major vendors like Cisco, NVIDIA, Celestica, Edgecore, and Wistron, offering hardware compatibility, support services, and enterprise-ready integrations.

Aviz Networks is actively enabling SONiC interoperability and provide full-stack support, automation, and performance optimizations—ensuring enterprises get the help they need.

Myth #4: SONiC is Not Cost-Effective

Reality: SONiC Delivers Up to 40% TCO Savings

Proprietary NOS vendors often claim that SONiC isn’t cost-effective when factoring in hardware, support, and integration costs. But PlugFest’s TCO analysis tells a different story:

  • Up to 40% lower TCO compared to proprietary NOS solutions
  • No NOS licensing costs – reducing OpEx significantly
  • Multi-vendor flexibility eliminates hardware markups and vendor lock-in

💰 Bottom line? SONiC saves enterprises millions while providing scalability for AI-driven workloads.

Myth #5: SONiC is Just a Trend—It Won’t Last

Reality: SONiC is the Future of Open Networking

Some skeptics believe SONiC is just another open-source experiment that will fade away. The reality? SONiC adoption is accelerating, with major enterprises, cloud providers, and AI data centers making it their NOS of choice.

With strong backing from industry leaders, ongoing community development, and real-world deployments, SONiC is shaping the future of open networking, AI infrastructure, and multi-vendor interoperability.

The Verdict: SONiC is Ready for Prime Time

SONiC NOS has matured beyond its hyperscaler origins into a battle-tested, cost-efficient, and enterprise-ready solution. As the 2025 PlugFest demonstrated, the myths surrounding SONiC are outdated, and the reality is clear:

  • Enterprise adoption is growing
  • Deployment is easier than ever
  • strong vendor ecosystem is supporting SONiC
  • TCO savings make it a smarter investment
  • Open networking is the future, and SONiC is leading the way

Frequently Asked Questions:

1. Is SONiC only suitable for hyperscalers like Microsoft?

No. While Microsoft pioneered SONiC, enterprises across industries such as healthcare, finance, retail, and AI-driven data centers are actively deploying it. The 2025 PlugFest validated SONiC for enterprise-grade use cases, including Layer 2 networking, AI fabric, and PoE-enabled whitebox switches, proving its viability beyond hyperscalers.

2. Is deploying SONiC difficult?

Not anymore. Early versions of SONiC required deep coding expertise, but today, automation and enterprise-ready solutions have simplified deployment. One-click SONiC migration guides and vendor support services ensure a seamless transition, offering the same ease of deployment as traditional NOS solutions.

3. Does SONiC lack vendor support?

No. The SONiC ecosystem is backed by major vendors, including Cisco, NVIDIA, Celestica, Edgecore, and Wistron. Aviz Networks provides full-stack SONiC support, automation, and performance optimization, ensuring enterprises have access to the help they need for successful adoption.

4. Is SONiC cost-effective compared to proprietary NOS solutions?

Yes. According to PlugFest’s TCO analysis, SONiC delivers up to 40% lower total cost of ownership (TCO) compared to proprietary NOS solutions. By eliminating NOS licensing costs, reducing OpEx, and providing multi-vendor flexibility, SONiC enables enterprises to save millions while ensuring scalability for AI-driven workloads.

5. Is SONiC just a passing trend?

No. SONiC is shaping the future of open networking. With increasing adoption by major enterprises, cloud providers, and AI data centers, SONiC is here to stay. Its strong backing from industry leaders and ongoing community development ensure long-term viability and continuous innovation.

6. How can enterprises migrate to SONiC?

Enterprises can transition to SONiC using migration guides, vendor-supported deployment tools, and automation frameworks. With one-click SONiC fabric deployment and extensive documentation, the migration process is streamlined for efficiency and minimal disruption.

7. What are the key benefits of adopting SONiC?

  • Vendor independence and flexibility
  • Lower TCO with no licensing costs
  • Enterprise-grade features and security
  • Strong industry support and ecosystem
  • Future-proof networking for AI and cloud workloads

8. Where can I learn more about real-world SONiC deployments?

Are you ready to break free from vendor lock-in and embrace open networking?
Learn More About SONiC and Migration Paths here.

Are you ready to break free from vendor lock-in and embrace open networking? Learn more about SONiC’s capabilities, migration paths, and real-world deployments
Book a Demo
Categories
Fabric Test Automation Suite

ONES 3.1 Boosts SONiC Support: Key Enhancements for Smarter Infrastructure Troubleshooting

In today’s fast-moving digital world, maintaining a stable and well-monitored infrastructure is crucial. The latest release of ONES 3.1 introduces key updates, including enhanced support for SONiC (Software for Open Networking in the Cloud). These enhancements boost visibility, automate critical processes, and strengthen system health monitoring. The improved SONiC support streamlines issue detection and response, optimizing performance and minimizing downtime. IT teams can now focus on strategic tasks, knowing their infrastructure is continuously and intelligently monitored for peak performance.
Stay ahead of issues and ensure smooth operations with ONES 3.1.

System Health Monitoring

CPU-Intensive Services

Previously, identifying resource-heavy processes was challenging due to the lack of granular insights in system-wide CPU and memory metrics. Often, system-level data shows a spike in CPU usage without providing a quick way to pinpoint the cause. To address this challenge, ONES now provides detailed reports on the top 10 CPU-consuming services running on the host, along with their memory usage. This helps users easily identify high-impact processes like redis-server, agent, syncd, and dockerd. With this level of detail, users can diagnose performance issues more quickly, optimize system resources, and prevent potential bottlenecks, resulting in greater system efficiency.
Top CPU consuming services

Unhealthy Devices with Failure Codes

ONES 3.1 introduces a new feature that highlights unhealthy devices, offering real-time failure detection for hardware (e.g., PSU, fan failures, LED alarms), software services, key processes, and containers. When a failure is detected, the device is marked as unhealthy, with detailed information readily available in the UI. This streamlined view helps operators quickly identify and resolve issues, simplifying troubleshooting. Notifications are also provided in the topology view and health summary page
Unhealthy Device with details of failure

SONiC Docker Transitions

Docker containers are the backbone of the SONiC operating system, and ensuring their stable operation is crucial for switch performance. Previously, tracking container state changes, such as shifts from “up” to “down,” was difficult and time-consuming. Operators often struggled to detect these changes in real-time, leading to delays in addressing service disruptions and unnoticed issues. ONES 3.1 introduces a new widget that visually highlights Docker container state transitions, allowing operators to quickly spot changes and respond to disruptions. Widgets provides a “Connect” button for direct SSH access to the switch, enabling swift action when needed. Additionally, it offers a timeframe selection feature, allowing operators to view container state changes over a specified period.
Docker transitions in the managed network

Automatic IP Detection, Alerting and Rediscovery:

When a monitored device’s management IP changes, it’s crucial for the monitoring software to update the IP promptly to ensure smooth operations. Previously, detecting and updating a device’s management IP was a manual, time-consuming process, often causing communication breakdowns and delayed issue identification. ONES 3.1 introduces an automatic rediscovery mechanism that instantly detects when a device’s management IP changes and re-registers the switch with the controller. This enhancement eliminates manual intervention, ensuring continuous communication, real-time monitoring, and faster issue resolution, even when devices are reconfigured.

Additionally, IP Transition Widget allows operators to track all IP changes the device has undergone over a specific period and if it had conflicted with another IP in the monitored network. To further enhance visibility, an alert generation option using ONES Rule engine notifies operators of any management IP changes, ensuring they are always aware of network modifications and can respond swiftly to maintain seamless operations.

Device IP transitions summary

Rule Engine: Enhanced Alerts

The ONES Rule Engine has emerged as a preferred tool for automating network monitoring, allowing operators to configure custom rules based on their specific threshold levels for various parameters. When a defined condition is met, the system automatically generates an alert, enabling real-time, proactive responses to potential issues. These new metrics provide deeper insights and more precise control over network performance, ensuring smoother operations and quicker issue resolution.

  • 1. Docker CPU/Memory Utilization – Enables container-level monitoring of CPU and memory usage, alerting operators if utilization exceeds set limits.
  • 2. Docker Down Status – Triggers an alert if any Docker container goes down, ensuring immediate awareness of service disruptions.
  • 3. Unhealthy Devices – Identifies and alerts operators of devices that experience failures.
  • 4. IP Transition Monitoring – Tracks IP changes in devices and generates alerts when IP transitions occur.

ONES 3.1 takes SONiC network monitoring and troubleshooting to the next level with powerful enhancements like real-time failure detection, automated IP rediscovery, detailed system health insights, and advanced alerting.
Ready to see ONES 3.1 in action? Book a demo today and experience how it can transform your network management with smarter automation and deeper insights.

Book a Demo
Categories
Fabric Test Automation Suite

Join Us at NVIDIA GTC: A Must-Watch Panel on AI in Networking!

AI is transforming every industry—including networking. As AI workloads scale, the infrastructure powering them must evolve. Networks must become smarter, faster, and more efficient to support the next wave of AI-driven innovation. At Aviz Networks, we believe in Networks for AI and AI for Networks—and we’re making it happen.

That’s why I’m thrilled to invite you to an exclusive panel at NVIDIA GTC, where we’ll explore NVIDIA’s role in transforming networking across these two dimensions and how Aviz Networks complements this ecosystem with innovative products.

Panel: Network Modernization in the Age of AI

AI-driven workloads demand a new era of networking—one that redefines how we design, deploy, and optimize infrastructure for peak performance. Our discussion will cover:

  • Networks for AI – Discover cutting-edge architectures designed to deliver high-performance, lossless networking for AI workloads
  • AI for Networks – See how AI-driven assistants like Network Copilot™ are redefining network operations with automation and real-time insights.
  • Aviz ONES + NVIDIA Spectrum™-X – Explore the next-gen AI networking solution built for Ethernet-based GPU cloud infrastructures.

Why This Matters

AI is no longer just an application—it’s the backbone of modern enterprise infrastructure. But here’s the challenge: AI workloads are hungry for bandwidth, require extreme precision, and demand real-time optimization. Aviz Networks and NVIDIA are solving this with cutting-edge AI networking innovations.

Don’t just read about AI in networking—experience it. Watch our exclusive demo and join the panel discussion at NVIDIA GTC!

Register Now

Contact Us

Join Us at NVIDIA GTC: A Must-Watch Panel on AI in Networking!

AI is transforming every industry—including networking. As AI workloads scale, the infrastructure powering them must evolve. Networks must become smarter, faster, and more efficient to support the next wave of AI-driven innovation. At Aviz Networks, we believe in Networks for AI and AI for Networks—and we’re making it happen. That’s why I’m thrilled to invite […]