The Tolly Group

🔍 Tolly Benchmark: Zyxel USG FLEX 100H vs. SonicWall TZ270 The Tolly Group put the Zyxel Networks North America USG FLEX 100H with the Gold Security UTM Bundle and the SonicWall TZ270 with the Advanced Protection Security Suite through a structured evaluation. Anti-malware detection, URL filtering accuracy, subscription pricing, and included features were all on the table. 📌 Here Is What Independent Testing Found 1️⃣ Zyxel returned higher anti-malware detection efficacy than SonicWall 2️⃣ URL filtering accuracy also favored the USG FLEX 100H 3️⃣ The subscription cost gap is hard to ignore: Zyxel runs 58% lower on one-year renewals and 61% lower on two-year renewals 4️⃣ Cloud management and lifetime firmware upgrades are part of the Gold Security UTM Bundle at no additional cost 📘 How Tolly Evaluated It Tolly engineers built a direct comparison using publicly available licensing data and documented feature sets from both vendors. 💡 Bottom Line: Stronger efficacy. Significantly lower renewal cost. More included. The data makes the case clearly. If your team is anywhere near a firewall decision, give the full report fifteen minutes. Read it here: https://lnkd.in/e4PZ-piQ #Firewall #NetworkSecurity #Zyxel #SonicWall #Cybersecurity

Tactical-edge compute doesn't get a climate-controlled rack. Anduril Industries commissioned The Tolly Group to test the Voyager VM 4.0 Tactical Edge Compute Platform through a 24-hour thermal chamber cycle: heat ramped up, held at sustained elevated temperatures, then brought back down. Sysbench, y-cruncher, and FIO ran every hour to check CPU, memory, and storage. The results: ✅ 0.22% Sysbench CPU variation ✅ ~1% combined CPU and memory variation ✅ ~3% storage read/write variation ✅ No degradation in any phase of the test When the mission moves beyond the data center, consistent performance is the baseline requirement. Read Tolly Report #226112: https://lnkd.in/eQj93_FX #EdgeComputing #TacticalEdge #Defense #Anduril #VoyagerVM #IndependentTesting #Tolly

When network and storage are monitored in separate tools, it takes longer to find the cause of a problem. In The Tolly Group's evaluation of Cisco + NetApp FlexPod, we validated integrated observability across the stack: ✔ Cisco Nexus Dashboard showed LAN and SAN fabrics together in one place ✔ Splunk Observability Cloud showed NetApp storage health, capacity, controller status, and IOPS live The result is one infrastructure with one view, so IT teams can spot issues sooner and spend less time switching between tools. Read the full report: https://lnkd.in/ekxR6s8H #FlexPod #Observability #Splunk #Cisco #NetApp #DataCenter #ITOps

PoE infrastructure keeps growing. Access points, IP cameras, phones, IoT devices — most of it running on 802.3bt at up to 90 watts per port. And most teams don't question compliance until something goes wrong in the field. That's a problem worth solving before deployment, not after. Kevin Tolly’s recent Analyst Preview covers the Sifos Technologies, Inc. MPD-600, a rack-mountable six-port PoE test instrument built specifically for that job. What it covers: 🔹 Six simultaneous test ports across the full power class range — Class 0 through Class 8 at up to 71W 🔹 Voltage testing at 43V, 49V, and 55VDC to simulate real cable run conditions 🔹 Automatic detection of inrush current violations, maintain power signature failures, and class-specific non-compliance 🔹 Three test modes: quick field verification, group interoperability testing, and extended diagnostic fault capture 🔹 Dedicated GbE ports for interacting with devices under active test 🔹 API support via PowerShell, Tcl/Tk, and C/C++/Python for production-line qualification IT teams, systems integrators, and device manufacturers all carry risk when a non-compliant device makes it into production. The MPD-600 is built to catch those failures on the bench, where fixing them is still straightforward. Watch the full analyst preview: https://lnkd.in/dZ7pfkPX #PoE #802dot3bt #NetworkCompliance #IoT #Sifos

GTM effectiveness is only as good as the platform behind it. We ran a real-world test to find out where Apollo.io stands. For our 2026 Go-To-Market Effectiveness and Data Quality Report, The Tolly Group evaluated Apollo.io by running an actual outbound campaign. Not a simulation. A live effort targeting 384 contacts at 205 companies over 30 days, with no email warm-up and no established sender reputation. 📈 2.37% cold-to-meeting conversion rate (industry avg: 0.5–1.5%) 📧 45% email open rate (industry standard: 27–40%) 🔗 One platform handling every GTM function end to end What We Evaluated: Contact data quality, email deliverability, AI-assisted workflows, and a competitive cost and feature comparison against Apollo's top alternatives. Apollo ranked as the most cost-effective vendor with the most complete standard plan. Bottom Line: Real-world testing under tough conditions produced results that cleared every benchmark we measured. Read the full report 👇 https://lnkd.in/dVyAQsDw #TollyGroup #GTM #SalesIntelligence #B2BSales #ColdEmail

🎥 This one keeps coming up in conversations about AI infrastructure. AI infrastructure is only as strong as the network underneath it. Is yours built for what's coming? The Ultra Ethernet Consortium is tackling that question head-on with open standards designed for the low-latency, lossless performance GPU clusters actually need. VIAVI Solutions’ Edward Nakamoto covers: 🔹 What makes AI networking fundamentally different 🔹 How UEC is eliminating vendor lock-in through open collaboration 🔹 The infrastructure requirements driving next-gen AI deployments 💡 Proprietary networking can't scale with AI indefinitely. UEC is building the foundation that can. Watch: https://lnkd.in/eAAB_7-e

We put the F5 BIG-IP Cloud-Native Edition head to head with a traditional VNF architecture, focusing on the workloads service providers care about most. The cloud-native approach came out ahead on latency and DNS throughput while holding steady on L4+CGNAT performance. For teams weighing a move away from VNFs, the webinar walks through what that migration actually looks like. Worth a watch. #ServiceProvider #CloudNative #NetworkModernization #F5 #IndependentTesting

Dell Technologies PowerSwitch SN5610 NVIDIA Spectrum with Dell SONiC AI Performance You can buy all the GPUs you want. If the network can't keep them fed, they wait. So we put Dell's AI Factory fabric through a simple question: how much of a 400GbE link do real AI workloads actually get to use? 📌 Key Findings 1️⃣ 392.18 Gbps of RDMA (RoCE) throughput, 98% of line rate 2️⃣ 389.84 Gbps of NCCL bandwidth on a 32 GPU AllReduce, the collective most common in AI training, at 97.5% efficiency 3️⃣ No measurable performance overhead from Dell SONiC, with enterprise features enabled 📘 What We Evaluated The fabric used Dell PowerSwitch NVIDIA Spectrum with Dell SONiC switches, with SN2201 switches for management. It connected Dell PowerEdge XE9680 servers with NVIDIA H200 SXM GPUs and BlueField-3 SuperNICs and DPUs. Two tests did the work: an RDMA microbenchmark for raw fabric bandwidth, and NCCL for real training communication. 💡 Bottom Line: For teams building on NVIDIA GPUs who want an open, Ethernet-based AI fabric, the results show it can meet both performance and operational requirements. Full report: https://lnkd.in/eAikkUC9 #AIInfrastructure #Networking #DellTechnologies #NVIDIA #SONiC

96% Accuracy. Zero Hallucinations. Independent Testing. Those aren't marketing numbers. They come from a Tolly benchmark that put octonomy Agentic AI up against Microsoft Copilot and a leading AI chatbot vendor using 50 questions pulled from a real, 1,000+ page industrial service manual. The questions were deliberately hard. Answers buried in engineering drawings. Values that required reading off a performance curve. Information scattered across separate document sections. The kind of complexity that exposes the difference between an AI that looks good in a demo and one that actually works in the field. 📌 Key Findings 1️⃣ octonomy Agentic AI: 96% accuracy — 48 of 50 questions answered correctly, no hallucinations 2️⃣ Microsoft Copilot (Direct Upload): 58% | Leading AI Chatbot Vendor: 34% | Microsoft Copilot (SharePoint): 26% 3️⃣ octonomy processed the full documentation library as a single unmodified knowledge base — competitors required manual pre-segmentation just to begin 📘 What Was Tested Four AI reasoning categories: precision data extraction from graphical sources, multi-document reasoning, visual and spatial interpretation, and complex structured data navigation. 💡 Bottom Line The gap between 96% and 26% is significant in any context. In technical, safety-critical, or high-stakes operational environments, it's the difference between a tool you can trust and one you can't. The Tolly Group report: https://lnkd.in/euQiKqsG #AgenticAI #AIAccuracy #EnterpriseAI #AIBenchmark #ThirdPartyValidation