How do I check if an IP address is blacklisted?

Use the free IP Blacklist & RBL Checker on Interconnected Systems Inc. It checks your IP against 17 validated DNS blacklists (SpamCop, Barracuda, DroneBL, UCEProtect, SpamEatingMonkey, and more) plus GreyNoise real-time threat intelligence, FireHOL Level 1, OTX AlienVault, and Emerging Threats — all in parallel. Results appear in about 3 seconds. A link to Cisco Talos Intelligence is included for full analysis.

What is an RBL or DNSBL check?

A Real-time Blackhole List (RBL) or DNS-based Blackhole List (DNSBL) is a database of IP addresses known to send spam or host malicious content. Mail servers query these lists to decide whether to accept email. You can check an IP against multiple RBLs simultaneously using the free tool at interconnectedsystems.net.

Can I check a URL for phishing or malware?

Yes — the URL Threat Checker on Interconnected Systems Inc. scans a URL or domain against the OpenPhish active phishing feed, GreyNoise Community, FireHOL Level 1, and 6 DNS blacklists on the resolved IP. It also provides direct links to Cisco Talos Intelligence and VirusTotal (70+ antivirus engines) for deeper analysis.

What is Cisco Talos Intelligence?

Cisco Talos is one of the world's largest commercial threat intelligence teams, responsible for the Snort IDS ruleset and the SpamCop spam-reporting service. Their reputation center (talosintelligence.com) provides email, web, and network threat context for IPs and domains. The IP reputation tool on this site links directly to Talos for full analysis.

What is a subnet calculator used for?

A subnet calculator helps network engineers divide an IP address range into smaller subnets, calculate network addresses, broadcast addresses, usable host ranges, and CIDR notation. The free IPv4 and IPv6 subnet calculators on Interconnected Systems Inc. also include wildcard mask conversion, CIDR quick reference tables, and IP range to CIDR block tools.

How can I do a BGP/ASN lookup?

The BGP & ASN tools on Interconnected Systems Inc. let you look up any Autonomous System Number (ASN) for prefix announcements, AS name, and registry data; look up which ASN and prefix owns any IP address; and access a live Route Views BGP terminal for real-time BGP table queries — all free, no login required.

How do I run a traceroute online without installing software?

Use the free Online Traceroute Tool at interconnectedsystems.net — no installation or login required. It runs a server-side traceroute from Toronto to any IPv4 or IPv6 address or hostname. Features include MPLS label stack decoding, ECMP path discovery (Paris traceroute style) to reveal load-balanced paths, NAT detection, Path MTU discovery, and ASN & geolocation enrichment per hop. Supports ICMP, UDP, and TCP probe protocols. Results appear in seconds.

What is MPLS traceroute and how does it work?

MPLS traceroute decodes the MPLS label stack carried in ICMP time-exceeded responses, revealing intermediate Label Switched Routers (LSRs) that are normally hidden inside a provider's MPLS core. The free traceroute tool at interconnectedsystems.net automatically decodes MPLS labels when present, showing the full forwarding path through carrier networks.

How do I compare AWS Direct Connect, Azure ExpressRoute, and GCP Cloud Interconnect costs?

Use the free Multi-Cloud Interconnect & Cost Planner at interconnectedsystems.net (Cloud Ops tab). Enter source and destination endpoints (AWS / Azure / GCP regions or on-premises), monthly data volume, sustained bandwidth, and latency target. The tool models monthly cost across IPsec VPN, AWS Direct Connect (1G/10G/100G hosted), Azure ExpressRoute (50 Mbps to 10 Gbps tiers), GCP Cloud Interconnect (Dedicated and Partner), PrivateLink/PSC, Transit Gateway / Virtual WAN / NCC, and SD-WAN — outputting a cost comparison table, latency range, setup time, reliability SLA, and a scored recommendation. All pricing data is sourced from official cloud provider APIs and published in /data/cloud-pricing.json for full transparency.

How do I check for VPC CIDR overlaps across multiple cloud accounts?

Use the free Multi-Cloud CIDR Planner & Overlap Detector at interconnectedsystems.net (Cloud Ops tab). Paste your CIDR blocks (one per line) and the tool runs pairwise overlap detection, returning a matrix of which blocks conflict. Reserved IPs are subtracted per cloud provider rules: AWS reserves 5 IPs per subnet (.0 network, .1 VPC router, .2 DNS, .3 future use, .255 broadcast); Azure reserves 5 (.0 network, .1 default gateway, .2-.3 Azure DNS, .255 broadcast); GCP reserves 2 (.0 network, .255 broadcast). The auto-allocate mode also carves non-overlapping subnets across AWS accounts, Azure VNets, and GCP VPCs from a single base CIDR.

How can I estimate AWS PrivateLink, Azure Private Endpoint, or GCP Private Service Connect costs?

Use the free Private Endpoint Cost & Sizing Calculator at interconnectedsystems.net (Cloud Ops tab). Enter the cloud provider, number of endpoints, average concurrent connections per endpoint, and monthly data processed. The tool calculates: AWS PrivateLink at $0.01/endpoint-hour + $0.01/GB; Azure Private Endpoint at $0.01/endpoint-hour + $0.01/GB inbound + $0.01/GB outbound; GCP PSC at $0.006/endpoint-hour + $0.01/GB. It also compares against equivalent public API egress to determine break-even data volume and recommends additional endpoints if concurrent connections exceed per-endpoint quotas (10,000 for AWS PrivateLink and GCP PSC, 1,000 for Azure Private Endpoint).

What is the cheapest way to transfer data between AWS, Azure, and GCP?

Inter-cloud egress through the public internet typically costs $0.08-0.09/GB across all three providers — making it the most expensive transfer category. Cheaper alternatives: (1) Co-locate AWS Direct Connect and GCP Dedicated Interconnect at the same colocation provider (e.g. Equinix) and transfer over private fiber at ~$0.02/GB; (2) Use a single primary cloud and PrivateLink/PSC for service exposure rather than replicating workloads cross-cloud; (3) Cache or CDN-offload large content to reduce origin egress. The free Egress & Data Transfer Cost Optimizer at interconnectedsystems.net (Cloud Ops tab) models all four traffic categories (internet, cross-region, inter-cloud, east-west) across providers with tiered pricing applied.

How do I detect shadow rules in Azure NSGs or GCP firewall rules?

Use the free Security Group / NSG / Firewall Rule Visualizer at interconnectedsystems.net (Cloud Ops tab). Paste your NSG rules from az network nsg show -o json or your firewall rules from gcloud compute firewall-rules list --format=json. The tool sorts rules by priority ascending and walks the list — for each rule, it checks every higher-priority predecessor to see if that earlier rule's source CIDR, destination port range, protocol, and direction fully cover the current rule's match space. If so, the current rule is shadowed and never executes. Both Azure NSGs (priority 100-4096) and GCP VPC firewalls (priority 0-65535) use first-match-wins evaluation, which makes shadow rules a common source of confusion when troubleshooting why a rule is not taking effect. The same tool also flags overly permissive rules (sensitive ports exposed to 0.0.0.0/0), least-privilege gaps (database ports public), and provides a what-if traffic tester.

How can I generate a Terraform AWS VPC, Azure Bicep VNet, or GCP firewall snippet?

Use the free Cloud Networking IaC Snippet Generator at interconnectedsystems.net (Cloud Ops tab). Pick the cloud (AWS, Azure, GCP), the IaC format (Terraform HCL, CloudFormation YAML, or Bicep), and a template (15 patterns covering VPC + subnets, Security Groups / NSGs / firewalls, peering, Transit Gateway, PrivateLink / Private Endpoint / PSC, etc.). Fill in the parameters — name, region, CIDRs, dependencies — and the tool generates a parameterized snippet validated for required fields, CIDR validity, CIDR overlap (subnet contained within VPC, sibling subnets non-overlapping), and basic syntax balance. Copy to clipboard or download as a .tf, .yaml, or .bicep file ready to apply with terraform apply, aws cloudformation deploy, or az deployment group create.

Certified Network Experts · Serving Enterprise Canada Since 2009

Internetworking Solutions

Delivered at the Speed of Light

From carrier-grade telecommunications to AI-driven data fabrics — we architect, deploy, and automate the networks that power the largest leading enterprises.

Explore Solutions Free Network Tools

20+Years Experience

ExpertCertified Engineers

100%Enterprise Focus

Who We Are

Design, Implementation, and Support — We've Got It Covered.

Interconnected Systems Inc. is a premium provider of enterprise and telecom internetworking consulting and solutions. We have decades of experience, and our clients include some of the biggest names in media and telecommunications.

Our experts are certified in the latest technologies and well-versed in tackling any challenge — whether you need solution design, infrastructure assessment, full-scale implementation, or around-the-clock troubleshooting. We engineer networks that are resilient, automated, and ready for the demands of AI-era workloads.

What We Do

End-to-End Network Services

From initial design through ongoing optimization — we deliver solutions at every stage of your network lifecycle.

Solution Design

Architecture that scales. We design multi-vendor network solutions — routed, switched, and overlay — aligned to your traffic patterns and business objectives.

L2/L3 Architecture
MPLS / SR
SD-WAN
DC Fabric

Infrastructure Assessment

Deep-dive audits of your existing environment. We identify performance bottlenecks, security gaps, and capacity constraints before they become outages.

Network Audit
Capacity Planning
Gap Analysis
Security Posture

Implementation & Deployment

Zero-downtime rollouts of complex network infrastructure. Certified engineers execute precision deployments across HPE, Juniper, and Cisco platforms.

Greenfield Builds
Migrations
Change Mgmt
Documentation

Network Automation

Transform operations with intent-based networking and infrastructure-as-code. Reduce toil, eliminate config drift, and deploy changes in minutes with Ansible and Python.

Ansible / NAPALM
Python / Nornir
CI/CD Pipelines
NetBox IPAM

AI Fabric & HPC Networking

Ultra-low-latency, high-throughput fabrics required for GPU clusters, ML training, and inference workloads. RDMA, RoCEv2, and lossless Ethernet expertise.

Spine-Leaf Fabric
RoCEv2 / RDMA
400G / 800G
Apstra IBN

NOC & Troubleshooting

When the network is down, every second counts. Rapid root-cause analysis and resolution for the most complex multi-vendor, multi-protocol environments.

24/7 Support
Protocol Debug
Packet Capture
Perf Tuning

Platform Expertise

Technologies We Master

Deep, vendor-certified expertise across the industry's leading networking platforms — from physical transport to cloud overlay.

CISCO

Cisco

Deep, hands-on expertise across IOS-XE, IOS-XR, NX-OS, ACI, SD-WAN, and Catalyst. Routing, switching, security, wireless, and data center — certified and field-proven.

IOS-XRNX-OS / ACISD-WANCatalystASR / NCS

JUNIPER

Juniper Networks

Junos OS experts delivering high-performance routing, switching, and security. Apstra-powered intent-based networking for modern AI-ready data centers.

Junos OSApstra IBNMX / QFX / EXSRXMist AI

HPE / ARUBA

HPE & Aruba

HPE ProLiant, Synergy, and Aruba CX switching for AI-ready infrastructure. AOS-CX, ClearPass NAC, and Central cloud management.

Aruba CXAOS-CXClearPassFlexFabricAruba Central

TELECOM

Telecommunications

Carrier-grade design and delivery. MPLS, BGP, optical transport, and last-mile connectivity for ISPs and large enterprise WAN environments.

BGP / MPLSDWDM / OpticalSR-MPLSSRv6EVPN

AUTOMATION

Network Automation

Infrastructure-as-Code for networking. Ansible, Python, YANG, RESTCONF/NETCONF, and GitOps pipelines that eliminate manual toil at scale.

AnsiblePython / NornirNETCONF / YANGTerraformGit / CI-CD

AI FABRIC

AI Fabric Design

Purpose-built fabrics for GPU-dense AI/ML infrastructure. Lossless RoCEv2, ECN, PFC, and congestion control optimized for distributed training workloads.

RoCEv2 / RDMA400G / 800GECMP / ECMP+ECN / PFCAI / ML Infra

AI · Cloud & Network Tools

Free AI Infrastructure, Cloud Infrastructure & Network Diagnostic Tools

AI infrastructure planning suite — LLM GPU & VRAM calculator, RoCEv2 fabric planner, inference serving simulator, vLLM/K8s config generator, TCO estimator, training storage sizer, and parallelism optimizer. Cloud Ops & multi-cloud networking — AWS/Azure/GCP interconnect & cost planner, VPC/VNet CIDR overlap detector, egress cost optimizer, PrivateLink/PSC sizing, hybrid connectivity comparator. Plus professional network tools: subnet calculator, BGP/ASN lookup, traceroute/MTR, DNS propagation, SPF/DMARC, VoIP, wireless calculators, and more. Free, no login required.

MCP Server — use every tool on this page from Claude or any MCP client

https://mcp.interconnectedsystems.net/mcp No token · no sign-in · Streamable HTTP · 19 tools

MAC Vendor Lookup

Identify the manufacturer of any network device from its MAC address using the full IEEE OUI registry (MA-L, MA-M, MA-S — 52,000+ entries, hosted locally).

My IP Address

Detect your public IPv4 and IPv6 addresses as seen from the internet.

IP Geolocation & ISP Info

Look up geolocation, ISP, ASN, and timezone for any IP address. Leave blank to look up your own IP.

Free IPv4 Subnet Calculator

Subdivide into /

CIDR ↔ Subnet Mask

Wildcard Mask Calculator

IPv6 Subnet Calculator

Subdivide into /

Network Path Traceroute / MTR

Trace the IPv4 or IPv6 network path from our server to any host — shows each hop, packet loss, and average latency. Full mode runs the trace for ~2 minutes with live-updating Sent / Received / Loss / Best / Avg / Worst / StDev statistics, just like running mtr in a terminal.

Advanced Path Trace (gtrace)

Deep network path analysis with automatic MPLS label-stack decoding, ECMP/load-balancing path discovery, NAT identification, Path MTU, and ASN/geo enrichment per hop. Multi-protocol: ICMP, UDP, or TCP.

Decode ICMP headers NAT detection ECMP discovery Path MTU

TCP Port Scanner

TCP connect scan from our server against any public host. Private IPs blocked. Max 20 ports, 30-second rate limit per IP. Custom: enter ports as 80,443,8000-8010.

Packet Header Size Reference

Protocol	Header (bytes)	Notes
Ethernet II	14	6 dst + 6 src + 2 EtherType
802.1Q VLAN	+4	TPID(2) + TCI(2) inserted
802.1ad Q-in-Q	+8	Outer + inner VLAN tags
IPv4	20 – 60	20 min; options add up to 40
IPv6	40	Fixed; ext. headers are separate
TCP	20 – 60	20 min; options (SACK, TS) up to 40
UDP	8	Fixed; src/dst port + len + checksum
ICMP / ICMPv6	8	Echo: type+code+checksum+id+seq
MPLS label	4 / label	Label(20b)+TC(3)+S(1)+TTL(8)
GRE	4 – 16	4 min; +4 each for key/seq/checksum
VXLAN	8	+outer UDP(8)+IP(20)+Eth(14) = 50 total
GENEVE	8+	8 fixed + variable TLV options
GTP-U	8 – 12	LTE/5G user plane; 8 min
IPsec ESP	8+	SPI(4)+Seq(4)+IV+payload+ICV
IPsec AH	24	Fixed for SHA-256; no encryption
NSH	8+	Network Svc Header (SFC path)
SRv6 SRH	8+(16×n)	8 base + 16 bytes per SID in list
PPPoE	+8	Over Ethernet; adds 6-byte hdr + 2 proto

Connection Quality Tester

Tests the connection from our server in Toronto, ON to your browser — measured independently over IPv4 and IPv6. Reports scored quality (0–100) with latency, jitter, throughput, and UDP reachability, plus a live map showing the path and approximate distance.

CIDR Quick Reference

CIDR	Subnet Mask	Hosts

TCP Throughput Calculator

Theoretical network limit via the Mathis formula: T = (MSS / RTT) × (1 / √Loss). Enter MSS, RTT, and observed loss rate to find the maximum achievable TCP throughput.

Real-World TCP Throughput

Bandwidth-Delay Product & TCP buffer sizing: BDP = BW × RTT, T ≤ buffer / RTT. Shows how much window you need and what throughput your current buffer actually delivers.

IP Reputation & Blacklist Check

Checks an IPv4 against GreyNoise, FireHOL Level 1 (Spamhaus DROP + DShield + Feodo), OTX AlienVault, Emerging Threats, and 17 DNS blacklists in parallel. Links to Cisco Talos Intelligence for full analysis.

URL / Domain Threat Check

Checks a URL or domain against OpenPhish (active phishing feed), GreyNoise, FireHOL Level 1, and 6 DNS blacklists on the resolved IP. Links to Cisco Talos and VirusTotal for deeper analysis.

AI Fabric Reference Library

Official validated design guides and reference architectures for AI GPU cluster networking. Click any document to open it in a new tab.

Juniper Networks — Validated Designs (JVD)

JVD

AI DC · Apstra + NVIDIA + WekaIO

Glittery Dawn — 3-stage Clos, rail-optimized GPU backend

JVD

AI DC · Apstra + AMD + Broadcom Thor2

3-stage Clos, Apstra-managed GPU fabric

JVD

AI DC · Apstra + AMD + VAST Storage

GPU backend fabric with VAST Data platform

JVD

GPU Backend · RDMA-LB + BGP-DPF

RDMA-aware load balancing, deterministic path forwarding

AMD — Instinct Reference Cluster Designs & Validated Architectures

AMD Instinct MI3XX Reference Design — 128 to 1024 GPU

MI300X / MI325X / MI350X / MI355X · generic BOM · fat-tree & rail Clos

AMD Instinct MI3XX Reference Design — 1024 to 8192 GPU

3-tier rail / scheduled fabrics · 51.2T design examples to 8K GPUs

NIC

AMD Pensando Pollara 400 — AI NIC architecture

400G AI NIC · multi-pathing · MRC · Vulcano 800 referenced for MI400

RACK

AMD Helios Rack-Scale (OCP ORW)

72× MI455X · UALoE scale-up · double-wide ORW liquid-cooled

PARTNER

Broadcom BCM957608 Ethernet Networking Guide — MI300X

Tomahawk + Pensando Pollara RoCEv2 config (Apr 2026)

PARTNER

Dell PowerEdge XE9680 + AMD MI300X — Multi-Node RoCE Training

Broadcom Thor 2 NICs · PowerSwitch Z9664F-ON · Dell-validated GPU fabric

PARTNER

HPE Helios AI Rack — MI455X + Broadcom Tomahawk 6

First commercial Helios deployment · UALoE scale-up + Ethernet (Dec 2025)

PARTNER

Cisco N9000 + AMD Pensando Pollara 400 Benchmark

Scale-out AI fabric validation · 128× MI300X · ROCm 6.3 / 7.0.3

NVIDIA — DGX SuperPOD Reference Architectures

DGX H100 SuperPOD

8× H100 per node, IB + Ethernet fabric, rail-optimized

DGX H200 SuperPOD

8× H200 per node, scalable unit design, IB fabric

DGX B200 SuperPOD

32× DGX B200 per SU, Blackwell, NVLink + IB fabric

DGX GB200 SuperPOD (NVL72)

8× DGX GB200 per SU, NVLink 5.0, Grace-Blackwell

DGX B300 SuperPOD · Spectrum-4

Spectrum-4 Ethernet + DC busbar, Blackwell Ultra

DGX B300 SuperPOD · Quantum-X800

NDR InfiniBand XDR fabric, Blackwell Ultra

Open Networking — Merchant Silicon Switch Platforms

OCP

Celestica DS5000 — OCP Specification

64× 800G OSFP · Broadcom Tomahawk 5 51.2T · 2U leaf/spine

Celestica DS5000 Product Page

Moonstone 800G platform — breakout profiles, BMC, timing

SPEC

Dell PowerSwitch Z9332F-ON

32× 400G QSFP-DD · Tomahawk 3 12.8T · 900 W typical

SPEC

Edgecore DCS240 / AS9726-32DB

32× 400G QSFP56-DD · Broadcom Trident 4 · 714 W max

SPEC

Edgecore DCS204 / AS7726-32X

32× 100G QSFP28 · Trident3-X7 3.2T · leaf and spine

OCP

Edgecore AS7326-56X — OCP Submission

48× 25G + 8× 100G top-of-rack · Trident3-X7 2.0T

ASIC

Broadcom Tomahawk 5 — 51.2T StrataXGS

The silicon behind every 64× 800G open platform

NOS

SONiC — config_db QoS Reference (PFC / ECN / DSCP)

PFC, ECN and DSCP mapping in config_db — the open-NOS reference

NVIDIA Quantum-X800 Platform

Q3400-RA: 144× 800G XDR over 72 OSFP cages · SHARP v4

GPU Fabric Topology Comparator — Rail-Optimized vs Standard

Visual comparison of three GPU cluster fabric architectures based on the Juniper JVD "Glittery Dawn" AI Data Center validated design and NVIDIA DGX SuperPOD reference. Shows rail-optimized (1 leaf per GPU rail), NVIDIA SuperPOD / GB200 NVL72, and standard fat-tree — with colored topology diagrams, PXN / NVLink annotations, and a side-by-side metric breakdown.

GPU servers

GPU / node type

Switch radix

LLM GPU & VRAM Calculator

Estimate VRAM requirements, KV cache, and maximum concurrent users for self-hosted LLM inference. Supports Llama, Qwen (including 3.8 27B), DeepSeek, Mistral, Gemma 4, gpt-oss, Kimi, GLM, Phi, and custom architectures. Multi-GPU tensor parallel and interconnect guidance included.

Model preset

Quantization

Context length (tokens)

Concurrent users

GQA / MLA KV efficiency factor

GPU model

GPU count (tensor parallel)

AI Fabric Planner — RoCEv2 GPU Cluster

Size and validate a production-grade RoCEv2 Ethernet fabric for GPU training and inference clusters. Models four parallelism strategies (Data / Tensor / Pipeline / Hybrid) — each with correct collective formulas, ring sizes, and fabric-score penalties. Uses the Juniper JVD stripe model for accurate leaf counts in rail-optimized designs. Outputs a full switch BOM with part numbers and cable estimates, 2-tier or 3-tier Clos (Superspine) topology diagrams, AllReduce / P2P transfer time, BDP / PFC / ECN thresholds, vendor config snippets, and a comprehensive CSV export. Includes RoCEv2 vs InfiniBand comparison.

GPU node preset

Total GPU count

Workload

Parallelism

Model / gradient size

Leaf / spine switch

Fabric topology

Oversubscription

AI Power, Cooling & Rack Planner

Calculate per-rack and cluster power draw (kW), cooling load (BTU / tons), PDU/UPS sizing, rack layout, and monthly OpEx for GPU clusters. Includes RoCE switch and NIC power — a cost factor most vendor tools omit. Chains from the Fabric Planner above.

GPU model

Total GPU count

GPUs / rack

Switch model (sets W / GPU below)

RoCE switch overhead (W / GPU-port) One switch port per GPU. A two-level network uses about three ports (the port facing the GPU, the uplink, and the spine), so switch power is about three times this number.

Cooling type

PUE (editable)

Electricity ($/kWh)

PDU rating (kW)

Power redundancy

AI TCO Estimator — On-Premise vs Cloud

Spreadsheet-style 3–5 year Total Cost of Ownership: CapEx breakdown, annual OpEx, break-even vs cloud GPU rental, and cumulative cost chart. Hardware costs pre-filled with 2025–2026 market pricing — edit any field to match your actual quotes.

GPU model

Planning horizon

Total GPU count

GPU unit cost ($)

Server chassis ($/node)

Networking total ($) A number you type is kept when the GPU count changes. Choosing a different GPU model fills it again.

Storage (TB) A number you type is kept when the GPU count changes. Choosing a different GPU model fills it again.

Storage ($/TB)

Install / integration (%)

Electricity ($/kWh)

PUE

Staff (FTE)

FTE cost ($/yr)

Maintenance (% hardware/yr)

Software licensing ($/yr)

Cloud comparison

Cloud rate ($/GPU-hr)

GPU utilization (%)

Training Data & Storage Sizer

Size NVMe/Ceph storage and I/O for LLM training. Computes dataset storage (with correct uint16/uint32 token precision for your vocabulary size), checkpoint size at 14 bytes/param (BF16 weights + FP32 master copy + Adam m+v — the full mixed-precision footprint), per-node ZeRO-3 checkpoint shards, data-loading I/O (correctly low — transformer LLMs are compute-bound, not I/O-bound), checkpoint burst bandwidth as the real storage sizing driver, and Ceph write-amplified east-west fabric impact. Filesystem tier recommendation is driven by checkpoint burst, not data-loading I/O. Ceph selected: shows OSD count, OSD server count, per-server RAM, and NIC sizing.

Dataset size (B tokens)

Token precision

Training epochs

Model size (B params)

Global batch size (tokens)

Checkpoints kept on disk

GPU nodes

Training duration (hours)

Storage replication factor

Checkpoint every N steps

ZeRO / FSDP sharding stage

Storage network

Parallelism Strategy & Scaling Recommender

Recommends the optimal Data / Tensor / Pipeline parallelism mix and ZeRO stage for your model and cluster. Uses exact ZeRO memory formulas from Rajbhandari et al. (SC'20): ZeRO-0 = 16Ψ, ZeRO-1 = 4Ψ + 12Ψ/Nd, ZeRO-3 = 16Ψ/Nd bytes per GPU. TFLOPS are dense BF16 (training uses dense arithmetic, not sparse). Activation memory is parameterized by sequence length. Pipeline bubble uses correct micro-batch count from seq_len. ZeRO-3 AllReduce overhead includes the 1.5× communication factor from the forward-pass AllGather. Directly informs 2-tier vs 3-tier fabric choice.

GPU model

Total GPU count

Model size (B params)

Target training time (hours)

Dataset size (B tokens)

Intra-node interconnect

Inter-node interconnect

Global batch size (tokens)

Sequence length (tokens)

Inference Serving Simulator

Simulate production LLM serving performance: TTFT, decode throughput, p95 latency, and max QPS for vLLM, TGI, Ollama, and TensorRT-LLM. Models continuous batching, paged-attention KV cache capacity, and multi-GPU tensor-parallel AllReduce east-west bandwidth.

Model

Quantization

GPU

GPU count (TP)

Serving framework

Prompt length (tokens)

Output length (tokens)

Request rate (req/s)

Max batch size

Target TTFT SLO (ms)

vLLM / Ollama / K8s Config Generator

Generate ready-to-copy deployment configs from your model and serving parameters: docker run, docker-compose.yaml, Kubernetes manifests, Helm values, and NCCL/RoCEv2 env vars. Full-deployment mode adds inline comments and production hardening; Skeleton outputs minimal working configs. Import directly from the Serving Simulator above.

Model

Serving framework

Quantization

GPU count (tensor-parallel-size)

Max model len (tokens)

Max concurrent (--max-num-seqs)

Output mode

Include multi-node NCCL / RoCEv2 env vars

Fabric switch config snippet

Cloud Provider Status

Live operational status for 21 cloud, AI, and networking providers — including Anthropic (Claude), OpenAI, Cohere, Together AI, Fireworks AI, Replit, Cerebras, and BaseTen. Data is fetched server-side and cached for 90 seconds.

ASN Lookup

Enter an AS number to retrieve BGP origin info, name, registry, country, and announced prefixes.

BGP Prefix Lookup

Look up BGP routing info for any IPv4 or IPv6 prefix — origin ASN, announcing organization, and upstream visibility.

IP BGP Origin Lookup

Enter any IP address to find its origin ASN, announced prefix, registry info, and geolocation.

Live Route Views Terminal

Interactive BGP terminal connected directly to public Route Views servers. Type Cisco IOS commands — try show ip bgp summary or show ip bgp 8.8.8.0/24. Read-only; no authentication required.

BGP Looking Glass Directory

Public BGP looking glasses — run traceroute, ping, and BGP table queries from global vantage points.

HE.netIPv4/IPv6 · 30+ PoPs BGP.toolsAdvanced BGP explorer RIPE StatRouting analytics HE BGP ToolkitASN / prefix / whois RPKI ValidatorCloudflare ROA check PeeringDBIX & peering data Lumen LGLevel3 / Lumen network Ping.peGlobal ping / MTR

DNS Record Lookup

Query any DNS record type via a public resolver of your choice. Select a DNS server from around the world to see what it returns.

Debug

Reverse DNS (PTR) Lookup

Resolve an IP address to its hostname via PTR record lookup. Works for both IPv4 and IPv6 addresses. Choose a DNS server to query.

DNS Propagation Checker

Compare DNS responses from Google (8.8.8.8) and Cloudflare (1.1.1.1) to verify a record has propagated globally.

Debug

Email Security (SPF / DMARC / MX)

Check a domain's email authentication setup — MX mail servers, SPF sender policy, and DMARC enforcement policy.

DNSSEC Validator

Check whether a domain has DNSSEC enabled and whether its signatures validate correctly via Google's validating resolver.

IP Blacklist / RBL Checker

Check an IPv4 address against 17 validated DNS blacklists (DNSBLs) in parallel. All sources verified for correct NXDOMAIN / A-record behaviour. Results in ~5 seconds.

Email Address Validator

Check syntax, MX records, SPF, DMARC, disposable domain detection, MX server reputation, and SMTP deliverability — all in one check. Results in ~5–10 seconds.

Disclaimers & how this works

No email is sent. The SMTP probe briefly connects to the mail server on port 25, sends EHLO and RCPT TO, then immediately disconnects. This leaves a single line in the target server's connection log.
Catch-all domains accept every address regardless of whether it exists. The probe detects this automatically and marks the result accordingly.
Major providers (Gmail, Outlook, Yahoo) may reject unknown addresses with a 550 error — but many also use greylisting or deferred rejection, making results inconclusive for individual addresses.
SMTP probe may be blocked by servers that require reverse DNS, allowlisted IPs, or TLS. In those cases the tool falls back to DNS-only analysis.
Results are informational only and reflect conditions at the time of the check. A "deliverable" result is not a guarantee of inbox delivery.

IPv4 ↔ Binary / Decimal

IPv6 Expander / Compressor

Bandwidth Calculator

IP Range → CIDR Blocks

Mist Toolbox — Chrome & Edge Extension

Read-only reporting for Juniper Mist and SSR, gathered into one browser extension. Pick a region, paste an Observer API token, choose a tool and press Run: the result previews on screen and lands as a styled .xlsx in your downloads. Everything runs in the browser — no Python, no server, no telemetry, not one third-party dependency.

Wireless

Disconnect Console — client-disconnect RCA: RF and SNR, 802.11 reason codes, DHCP after roam, RRM occupancy, 7-day DFS radar, Teams / Zoom call quality
Client Wi-Fi PHY Inspector — one client graded on RSSI, SNR, noise, PHY rate vs ceiling, retries, channel load and co-channel APs
SSID Report — every SSID per site, including org-template WLANs, with auth, VLAN and bands
Wi-Fi Clients Export — every connected client with RF stats, 802.11 standard and guest-portal fields

Switching

Switch Port Inventory — every port on one sheet: 41 columns of live status, PoE, LLDP and STP merged with the configured profile
Switch Software Report — model, serial, version, uptime and per-member Virtual Chassis versions
Switch Config Export — running config as Junos set commands, one .zip with a folder per site
Switch PSU Status — failed supplies and switches without PSU redundancy flagged
Switch Additional CLI — template, site and device-level CLI side by side
IP Blocks / IRB Report — every IRB address with subnet math, duplicate and overlapping networks flagged across sites

Routing, WAN & org

BGP Sessions — switch EVPN and underlay, WAN-edge BGP and every SSR / SRX peer path in one org-wide pass, problems first on a colour board
SSR Pre/Post Check — snapshot BGP, OSPF, interfaces, adjacencies, sessions and alarms on a Conductor, change, then diff every value that moved
Site Alarms — every alarm across one site or the whole org for the last day or week

Extend it with any AI assistant

The 13 built-in tools are a starting point. A tool is one JavaScript file; the toolbox supplies the menu card, token and org handling, the site picker, the settings form, progress, preview and the .xlsx. The extension ships its own LLM spec, TOOL_PROMPT.md — press Manage tools → Copy AI prompt, paste it into Claude, ChatGPT, Copilot or Gemini, then paste a Python or Node script you already have, or just describe the report you want.

your script or idea  +  TOOL_PROMPT.md  →  AI assistant  →  my-report.js  →  Manage tools → Install  →  new tool card

The returned file is checked before anything is written: Node or Python code, raw fetch, storage, console.*, POSTs and npm imports are refused with an explanation of what to fix. Once it passes, it is written into the extension’s tools/ folder and its card appears without a reload, so your token survives — with a Remove button beside it. The check catches mistakes, not malice: install tools only from people you trust.

The token lives in memory in that one tab — never written to storage, never sent to the service worker, never logged — and is wiped after 30 idle minutes or when the tab closes. Nothing issues a write to Mist. 13 pinned Mist API hosts and no permissions entries; the SSR Conductor origin is asked for once, at runtime, the first time you connect.

Every tool

The Excel workbooks they save

Add your own tool

Latest release v1.1.1 · 13 built-in tools · MIT licence · captures use synthetic data from an 8-site demo org

Download mist-toolbox.zip Source & releases on GitHub

RF Power Converter

Convert between dBm, dBW, mW, W, and µW. Essential for reading RF specs and regulatory limits.

Free Space Path Loss (FSPL)

Calculate signal attenuation over distance using the Friis transmission equation. Ideal conditions, no obstacles.

EIRP Calculator

Effective Isotropic Radiated Power — Tx Power + Antenna Gain − Cable Loss. Used for regulatory compliance.

Link Budget Calculator

Calculate received signal power and link margin for a point-to-point wireless link.

WiFi Signal Quality

Translate a received signal level (dBm) into quality rating, estimated SNR, and expected 802.11ax throughput.

Wi-Fi Channel Reference

Ch	Center (MHz)	Notes

Codec Bandwidth Calculator

Calculate bandwidth per call including RTP/UDP/IP and Ethernet overhead for common VoIP codecs.

VoIP Call Capacity

How many concurrent VoIP calls fit in a given link bandwidth? Results shown per codec (full-duplex, with Ethernet overhead).

MOS / Voice Quality (E-Model)

Estimate MOS (1–5) and R-Factor using the ITU-T E-Model. Accounts for codec, one-way delay, packet loss, and jitter.

DTMF Tone Generator

Click keys to generate DTMF tones via the Web Audio API. Plays dual-tone multi-frequency signals used in telephony.

SIP Server DNS Lookup

Look up SIP SRV records (_sip._udp, _sip._tcp, _sips._tcp) to find SIP server addresses and ports for a domain.

SIP Response Code Reference

Code	Description

RFC Daily Brief

Daily operator brief of new IETF RFCs, in-scope Internet-Drafts, and protocol-registry movement — networking, routing, ops, transport, VoIP/IMS, and protocol security. Analysed with Grok 4.7. RFC Editor ↗ · Datatracker ↗

Pricing Data & Sources

Updated … View JSON ↗ Export all values ↓

Every cost calculation in this tab uses rates from the official cloud provider pricing APIs, refreshed daily. The full pricing dataset is published as a static JSON file so you can audit every number used — no hidden multipliers or undocumented assumptions.

Multi-Cloud Interconnect & Cost Planner

Select source and destination cloud regions or on-premises, define bandwidth and latency requirements, and compare all connectivity options — VPN, Direct Connect/ExpressRoute/Cloud Interconnect, PrivateLink/PSC, Transit Gateway/Virtual WAN/NCC, and SD-WAN — with monthly cost estimates, latency ranges, setup timelines, and a recommendation.

Source provider

Source region

Destination provider

Destination region

Monthly data (GB)

Sustained BW (Mbps)

Latency target

Connectivity type

Advanced — data flow pattern (optional)

Defaults assume constant 24/7 outbound traffic. If your monitoring shows a different pattern, dial it in here for a more accurate BOM.

Workload preset

Outbound % of monthly data 100%

Peak / sustained ratio 1.0×

Active hours / day 24 h

Compression ratio 1.0×

HA redundancy (gateways) 2

Reservation discount 0%

Monthly growth rate 0% / mo

Multi-Cloud VPC/VNet/VCN CIDR Planner & Overlap Detector

Detect overlapping CIDR blocks across VPCs, VNets, and VCNs — or auto-allocate non-overlapping subnets for a multi-cloud environment. Applies cloud-specific reserved IP rules: AWS reserves 5 IPs per subnet (.0 network, .1 router, .2 DNS, .3 future use, .255 broadcast); Azure reserves 5 (.0 network, .1 gateway, .2–.3 DNS, .255 broadcast); GCP reserves 2 (.0 network, .255 broadcast).

Detect Overlaps Auto-Allocate

CIDR blocks (one per line)

Cloud (for reserved-IP annotation)

Enter any mix of IPv4 CIDRs. Results show pairwise overlap detection and the usable host count per subnet after cloud-specific IP reservations.

Egress & Data Transfer Cost Optimizer

Drag the sliders to define monthly traffic volumes across four cost categories. Compare total egress spend across AWS, Azure, and GCP with tiered pricing applied. Get actionable tips to reduce spend via PrivateLink, CDN placement, or traffic routing changes.

Cloud providers

AWS Azure GCP OCI

Primary region tier

Internet egress (GB/mo)

10,000 GB

Cross-region transfer (GB/mo)

2,000 GB

Inter-cloud transfer (GB/mo)

1,000 GB

Intra-region east-west (GB/mo)

5,000 GB

Private Endpoint / Service Connect Cost & Sizing

Estimate monthly costs for AWS PrivateLink, Azure Private Endpoints, or GCP Private Service Connect. Model endpoint hours, data processing charges, and compare against public API egress costs. Includes auto-scaling guidance based on concurrent connection load.

Cloud provider

# of endpoints

AZs (AWS only)

Avg concurrent connections per endpoint

Monthly data processed (GB)

Equivalent public API traffic (GB/mo)

Compare vs public endpoint cost

Hybrid Connectivity Options Comparator

Define your on-premises location, target cloud, bandwidth requirements, latency SLA, and compliance needs. Get a side-by-side comparison of all hybrid connectivity options — IPsec VPN, dedicated circuits (Direct Connect / ExpressRoute / Cloud Interconnect), PrivateLink/PSC, Transit Gateway, and SD-WAN — with monthly costs, failure modes, and a scored recommendation.

On-premises location

Cloud provider

Required BW (Mbps)

Monthly data transfer (GB)

Latency requirement

Primary use case

Compliance requirements

PCI-DSS HIPAA SOC 2

Security Group / NSG / Firewall Rule Visualizer & Conflict Analyzer

Paste rules from AWS describe-security-groups, Azure az network nsg show, or GCP gcloud firewall-rules list — or build them in the table editor. Detects shadow rules (Azure/GCP first-match-wins), redundant rules (AWS), overly permissive entries (sensitive ports exposed publicly), and least-privilege gaps. Includes a what-if traffic tester that walks the ruleset to predict allow/deny verdicts.

Cloud platform

Rule entry mode

Build in table Paste CLI / IaC output

Demo rulesets

What-if traffic tester — walk the current ruleset to predict the verdict for one specific flow

Source IP

Protocol

Dest port

Direction

Cloud Networking IaC Snippet Generator

Generate validated Terraform, CloudFormation, or Bicep snippets for common multi-cloud networking patterns — VPCs/VNets, subnets, security groups/NSGs/firewalls, peering, Transit Gateway/Virtual WAN, PrivateLink/PSC, and more. Each snippet is parameterized; outputs are checked for required fields, CIDR validity, CIDR overlap (subnet within VPC), and basic syntax balance. Copy or download as a ready-to-apply file.

Cloud

IaC format

Template

Administrative Distance / Route Preference

Cisco & Arista use Administrative Distance (0–255). Juniper uses Route Preference (same semantics, wider scale). Lower value wins on all platforms.

Protocol / Source	Cisco IOS/XE/XR	Juniper Junos	Arista EOS
Connected	0	0	0
Static	1	5	1
EIGRP Summary	5	—	—
EIGRP Internal	90	—	—
OSPF Internal	110	10	110
IS-IS L1 Internal	115	15	115
IS-IS L2 Internal	115	18	115
RIP	120	100	120
OSPF External	110	150	110
IS-IS L1 External	115	160	—
IS-IS L2 External	115	165	—
EIGRP External	170	—	—
iBGP	200	170	200
eBGP	20	170	200 ¹

¹ Arista defaults eBGP and iBGP both to AD 200. Use distance bgp 20 200 200 for Cisco-parity. Junos uses the same preference (170) for both iBGP and eBGP. EIGRP is Cisco-only.

OSPF

LSA Types

Type	Name	Scope
1	Router LSA	Intra-area
2	Network LSA (DR)	Intra-area
3	Summary Network LSA (ABR)	Inter-area
4	ASBR Summary LSA (ABR)	Inter-area
5	AS External LSA (ASBR)	AS-wide
7	NSSA External LSA	NSSA area
9	Opaque (link-local, TE, SR)	Link-local
10	Opaque (area, TE RSVP)	Intra-area
11	Opaque (AS-wide)	AS-wide

Neighbor States

Down → Attempt → Init → 2-Way → ExStart → Exchange → Loading → Full

Area Types

Area	LSA 5	LSA 3/4	LSA 7
Backbone (0)	✓	✓	✗
Stub	✗	✓	✗
Totally Stub	✗	✗ (default only)	✗
NSSA	✗	✓	✓
Totally NSSA	✗	✗	✓

Key timers: Hello 10 s / Dead 40 s (NBMA: 30/120). LSA max-age 3600 s. DR/BDR elected on broadcast & NBMA only.

Vendor Differences

Reference BW	Default 100 Mbps on all three. Tune with `auto-cost reference-bandwidth` (Cisco/Arista) or `reference-bandwidth` (Junos).
Dead/Hello sync	Arista: dead-interval is not auto-adjusted when hello-interval changes — set both manually. Cisco & Junos auto-derive dead = hello × 4.
GR helper mode	Juniper: enabled by default (RFC 3623). Cisco & Arista: require explicit configuration.
External routes	E2 is default on all three. E1 cost = redistributed metric + cost to ASBR. When both E1 and E2 exist for same prefix, E1 is always preferred.

BGP — iBGP vs eBGP

Property	iBGP	eBGP
AD	200	20
NEXT_HOP	Unchanged	Set to self
AS_PATH	Unchanged	Prepend own ASN
LOCAL_PREF	Propagated	Not sent out
MED	Not propagated	Sent to eBGP peer
Full mesh	Required (or RR/Confed)	N/A
TTL	255	1 (use EBGP multihop)

Path Selection (in order)

1	Highest WEIGHT (Cisco-local)
2	Highest LOCAL_PREF
3	Locally originated (network/aggregate)
4	Shortest AS_PATH
5	Lowest ORIGIN (IGP < EGP < Incomplete)
6	Lowest MED (same AS)
7	Prefer eBGP over iBGP
8	Lowest IGP metric to NEXT_HOP
9	Oldest eBGP path / lowest Router ID
10	Lowest peer IP address

Path Selection — Vendor Differences

Cisco IOS/XE/XR	Step 1 is WEIGHT (local, not advertised; default 0 learned / 32768 local). Remaining steps as listed above.
Juniper Junos	No WEIGHT — selection begins at LOCAL_PREF. Next-hop must be resolvable (implicit prereq). Final tiebreaker: lowest router ID.
Arista EOS	WEIGHT supported (Cisco-compatible step 1). Final tiebreaker: lowest peer IP address.

Tune with bgp bestpath knobs (Cisco/Arista) or set protocols bgp path-selection (Junos).

IS-IS

Levels

Level	Function	LSDB
L1	Intra-area routing	Area LSDB
L2	Inter-area (backbone)	Backbone LSDB
L1/L2	ABR equivalent	Both LSDBs

PDU Types

PDU	Purpose
IIH	IS-IS Hello — neighbour discovery & keepalive
LSP	Link State PDU — topology advertisement
CSNP	Complete Seq. — full LSDB summary (DR only)
PSNP	Partial Seq. — request / acknowledge LSPs

Key TLVs

TLV	Description
1	Area Addresses
2	IS Neighbours (L1)
22	Extended IS Reachability (TE metric)
128	IP Internal Reachability
130	IP External Reachability
135	Extended IP Reachability (TE)
232	IPv6 Reachability
236	IPv6 Interface Address

NET format: 49.0001.0100.1001.0001.00 — Area · System-ID · NSEL(00)

Vendor Differences

Metric style	Cisco: narrow default — add `metric-style wide` for TE/SR. Juniper: advertises both narrow + wide by default. Arista: narrow default.
DIS hello timers	Juniper: DIS hello 3 s / hold 9 s (÷3 from standard 9 s/27 s). Cisco: does not auto-scale DIS hellos.
Interface default metric	10 on all three platforms.

MPLS

Label Operations

Op	Description
PUSH	Add label (ingress LER)
SWAP	Replace label (transit LSR)
POP	Remove label (egress LER)
PHP	Pop one hop before egress. The egress signals label 3 (implicit null)

Label Stack Entry (32 bits)

Label (20b) · TC/EXP (3b, QoS) · S bit (1b, bottom-of-stack) · TTL (8b). 4 bytes per label.

Special-Purpose Labels (0–15)

Label	Meaning
0	IPv4 Explicit NULL. Keeps TC bits to the egress
1	Router Alert
2	IPv6 Explicit NULL
3	Implicit NULL. Signalled only, never on the wire
4	MPLS Network Actions (MNA), RFC 9994
7	Entropy Label Indicator (ELI), RFC 6790
13	Generic Associated Channel Label (GAL)
14	OAM Alert
15	Extension Label (XL). Next label is an extended special-purpose value

Labels 5, 6 and 8–12 are unassigned. Ordinary labels start at 16.

Entropy Label

Ingress pushes ELI (7) + EL, 8 bytes. Transit LSRs hash on the EL for ECMP without parsing the payload. The egress advertises Entropy Label Capability (ELC). Each LSR advertises its Entropy Readable Label Depth (ERLD), which limits how deep the ELI/EL pair may sit in an SR label stack (RFC 8662).

MPLS Network Actions (MNA)

New in 2025–2026. A sub-stack that starts with label 4 carries network actions and their data inside the label stack (framework RFC 9789, encoding RFC 9994, June 2026). Use cases are in RFC 9791. Carrying data after the stack (post-stack header) is still an Internet-Draft: draft-ietf-mpls-mna-ps-hdr-19, in the RFC Editor queue as of Nov 2026.

SR-MPLS

SID Types

SID	Description
Prefix-SID	Global. An index into the SRGB: label = SRGB base + index
Node-SID	Prefix-SID on a node loopback (N flag set)
Anycast-SID	Same Prefix-SID advertised by several nodes
Adj-SID	Local to one node. One per adjacency, dynamic or from the SRLB
Binding SID	Local label that steers into an SR Policy
BGP Prefix-SID	Prefix-SID carried in BGP labeled unicast (RFC 8669)
Flex-Algo SID	Extra Prefix-SID per algorithm 128–255 (RFC 9350)

Algorithm 0 = SPF, 1 = strict SPF, 128–255 = operator-defined Flex-Algo (metric type + constraints).

Blocks and Limits

SRGB	Global block for Prefix-SIDs. Use the same range on every node
SRLB	Local block for Adj-SIDs and other local SIDs
MSD	Maximum SID Depth a node can push. Advertised in IS-IS (RFC 8491) and OSPF (RFC 8476)
P / E flags	P = no PHP. E = replace with explicit null

SR Policy (RFC 9256)

Identified by (headend, color, endpoint). Holds candidate paths ranked by preference; each has one or more weighted segment lists. A Binding SID represents the policy. Signalled by BGP (RFC 9830), PCEP (RFC 8664) or local config. Service routes steer in by matching the BGP color community.

Fast Reroute

TI-LFA (RFC 9855, Oct 2025): the backup follows the post-convergence path, encoded as a short repair segment list. Covers link, node and SRLG failures in any topology.

Vendor Differences

SRGB (Cisco)	16000–23999 default. Configure with `segment-routing global-block`.
SRGB (Juniper)	No fixed default. Taken from the dynamic label pool, so set `srgb start-label` explicitly.
SRGB (Arista)	900000–965535 default. Configure with `segment-routing global-block`.
TI-LFA	Cisco: IS-IS and OSPF. Juniper: IS-IS and OSPFv2. Arista: IS-IS documented; check release notes for OSPF.

With different SRGBs per vendor, the same index gives a different label on each node. Align the SRGB everywhere, or plan SIDs as indexes, never as absolute labels.

SRv6 & uSID

SID Structure

An SRv6 SID is an IPv6 address: Locator : Function : Argument, 128 bits in total. The locator is Block : Node and is advertised as an ordinary IPv6 prefix, so nodes that are not segment endpoints only need plain IPv6 forwarding.

Segment Routing Header (RFC 8754)

IPv6 Routing header type 4. 8 bytes fixed + 16 bytes per SID. The segment list is stored in reverse order; Segments Left points at the active SID, which is also the IPv6 destination address.

Endpoint Behaviours (RFC 8986)

Behaviour	Description
End	Node segment. SRv6 form of a Prefix-SID
End.X	L3 cross-connect to one adjacency. SRv6 form of an Adj-SID
End.T	Lookup in a specific IPv6 table
End.DT4 / DT6 / DT46	Decapsulate, then look up in a VRF table (per-VRF L3VPN)
End.DX4 / DX6	Decapsulate, then cross-connect to a CE adjacency (per-CE L3VPN)
End.DX2	Decapsulate, then send the Ethernet frame out an interface (EVPN VPWS)
End.DT2U / DT2M	Decapsulate, then bridge: unicast MAC lookup / BUM flood (EVPN)
End.B6.Encaps	Binding SID into an SRv6 Policy
End.BM	Binding SID into an SR-MPLS Policy

Flavors and Headend

PSP	Penultimate Segment Pop. SRH removed one segment early
USP	Ultimate Segment Pop. SRH removed at the last segment
USD	Ultimate Segment Decapsulation. Outer IPv6 header removed at the last segment
H.Encaps	Headend adds an outer IPv6 header + SRH
H.Encaps.Red	Same, but the first SID is only in the destination address. With one segment the SRH can be left out

uSID / Compressed SID (RFC 9800, June 2025)

NEXT-CSID	The uSID flavor. Several short SIDs packed into one 128-bit container
F3216	32-bit Locator-Block + 16-bit uSIDs. The format every NEXT-CSID implementation must support
Capacity	(128 − 32) / 16 = 6 uSIDs per container
Forwarding	The node that owns the first uSID shifts the rest left by 16 bits, zero-fills the end, and forwards on the new destination address
GIB / LIB	Global ID Block for domain-wide uSIDs (nodes). Local ID Block for per-node uSIDs (adjacencies, services). The split is a deployment choice
REPLACE-CSID	The other RFC 9800 flavor. 32-bit CSIDs packed in the SRH replace part of the destination address at each hop

Example with block 2001:db8::/32: 2001:db8:100:200:300:: → node 0100 → 2001:db8:200:300:: → node 0200 → 2001:db8:300::. Three segments, no SRH.

Vendor shorthand, not RFC terms: uN = End with NEXT-CSID · uA = End.X with NEXT-CSID · uDT4 / uDT6 / uDT46 = End.DT4 / DT6 / DT46 · uDX2 = End.DX2. Decapsulation SIDs sit last in the list, so they need no NEXT-CSID flavor.

Addressing and Control Plane

IANA assigned 5f00::/16 for SRv6 SIDs (RFC 9602); it is not globally reachable and is easy to filter at the domain edge. Locators are flooded by IS-IS (RFC 9352) or OSPFv3 (RFC 9513). VPN and EVPN service SIDs are carried in BGP (RFC 9252). Ping and traceroute to SIDs: RFC 9259.

SR-MPLS vs SRv6 vs uSID

Property	SR-MPLS	SRv6 (full SID)	SRv6 uSID
Data plane	MPLS label stack	IPv6 + SRH	IPv6. SRH only beyond 6 uSIDs
Segment size	4 bytes	16 bytes	2 bytes (F3216)
1 segment	4 B	40 B	40 B
5 segments	20 B	112 B	40 B
10 segments	40 B	192 B	64 B
Transit nodes	Must forward MPLS	Plain IPv6 is enough	Plain IPv6 is enough
ECMP entropy	Entropy label (+8 B)	IPv6 flow label	IPv6 flow label
Summarization	No. One label per prefix	Yes. Locators aggregate	Yes. Locators aggregate
Service demux	VPN label	End.DT / End.DX SID	Service uSID, last in list
Fast reroute	TI-LFA	TI-LFA	TI-LFA

Overhead is the added encapsulation with reduced encapsulation (H.Encaps.Red): 40 B outer IPv6 header, plus an SRH of 8 B + 16 B per remaining SID or uSID container. Count the service SID as a segment. Budget MTU for it on every core link.

Interworking between SRv6 and MPLS domains is still an Internet-Draft: draft-ietf-spring-srv6-mpls-interworking-03 (Aug 2026).

Standards Map (checked Nov 2026)

Document	Subject
RFC 8402	Segment Routing architecture
RFC 8660	SR with the MPLS data plane
RFC 8754	IPv6 Segment Routing Header
RFC 8986	SRv6 network programming (behaviours)
RFC 9252	BGP VPN and EVPN services over SRv6
RFC 9256	SR Policy architecture
RFC 9350	IGP Flexible Algorithm
RFC 9602	SRv6 SID address block 5f00::/16
RFC 9800	Compressed SRv6 segment lists (uSID), June 2025
RFC 9830	SR Policies in BGP, Sept 2025
RFC 9855	TI-LFA, Oct 2025
RFC 9871	BGP Color-Aware Routing (Experimental), Nov 2025
RFC 9832	BGP Classful Transport (Experimental), Sept 2025
RFC 9994	MPLS Network Actions sub-stack, June 2026
srv6-security-19	Draft. SRv6 security considerations. RFC Editor queue
srv6addressing-01	Draft. SRv6 addressing recommendations

Drafts change and can expire. The RFC Brief tab tracks new SPRING, MPLS and SRv6 documents daily.

EVPN Route Types

Type	Name	Purpose
1	Ethernet Auto-Discovery	Mass withdrawal, aliasing (multi-homing)
2	MAC/IP Advertisement	MAC learning, ARP suppression, host routes
3	Inclusive Multicast ET (IMET)	BUM flooding — announce VTEP membership
4	Ethernet Segment Route	DF election, multi-homing ES discovery
5	IP Prefix Route	IRB/inter-subnet routing, L3VPN prefixes

VXLAN Concepts

Term	Detail
VNI	24-bit VXLAN Network ID (~16M segments)
UDP port	4789 (IANA) — outer UDP encapsulation
VTEP	VXLAN Tunnel Endpoint — encap/decap point
L2 VNI	Bridge domain / broadcast domain
L3 VNI	VRF / inter-subnet routing (symmetric IRB)

IRB Models

Model	Route at	L3 VNI
Asymmetric	Ingress only	Not required
Symmetric	Ingress + Egress	Required (per VRF)

RT format: ASN:VNI. BGP EVPN AFI 25/SAFI 70. Type-2 carries MAC+IP, enables ARP suppression. Type-5 enables host-route injection from VMs.

BGP Attributes Quick Reference

Attribute	Type	Scope
ORIGIN	Well-known mandatory	IGP / EGP / incomplete
AS_PATH	Well-known mandatory	Loop prevention + path selection
NEXT_HOP	Well-known mandatory	Next-hop IP for prefix
LOCAL_PREF	Well-known discretionary	iBGP only — exit selection
ATOMIC_AGGR	Well-known discretionary	Aggregation indicator
MED	Optional non-transitive	Suggest entry point to neighbour AS
COMMUNITY	Optional transitive	32-bit tag — policy signalling
EXT_COMMUNITY	Optional transitive	64-bit — VPN RT/RD, SoO, CoS
LARGE_COMMUNITY	Optional transitive	96-bit — RFC 8092
AGGREGATOR	Optional transitive	ASN + IP of aggregating router

Well-known Communities

Value	Meaning
NO_EXPORT (0xFFFFFF01)	Do not advertise beyond AS boundary
NO_ADVERTISE (0xFFFFFF02)	Do not advertise to any peer
NO_EXPORT_SUBCONFED (0xFFFFFF03)	Do not export to eBGP sub-confederation peers
BLACKHOLE (65535:666)	RTBH — discard matching traffic (RFC 7999)

Common BFD Timers (reference)

Min interval 50–300 ms typical. Multiplier 3 (detect = interval × mult). Async mode standard; Echo mode for single-hop. Micro BFD for LAG members.

BGP & Routing

Hurricane Electric BGP ToolkitASN, prefix, trace, whois
HE.net Looking GlassTraceroute / ping from 30+ global PoPs
BGP.toolsAdvanced BGP explorer & visualizer
RIPE StatBGP analytics, routing history, visibility
RPKI Validator (Cloudflare)Validate ROAs and route origin authorization
PeeringDBIX and network peering database
BGPlayAnimated BGP routing history visualization
Route ViewsBGP routing table archive

Network Automation

Ansible Network AutomationAgentless multi-vendor automation
NAPALMMulti-vendor network API library
NornirPython-native automation framework
NetmikoSSH library for network devices
BatfishNetwork configuration analysis and verification
TerraformInfrastructure as Code
NetBoxIPAM, DCIM & network source of truth

Monitoring & Observability

PrometheusOpen-source metrics & alerting
GrafanaNetwork dashboards & visualization
LibreNMSAuto-discover SNMP/ICMP monitoring
ntopngReal-time network traffic analysis
ZabbixEnterprise-grade monitoring platform
ELK StackLog aggregation for syslog & NetFlow

Packet Analysis & Testing

WiresharkThe definitive packet protocol analyzer
tcpdumpCLI packet capture and analysis
iPerf3Network throughput measurement
NmapNetwork discovery and security scanning
Ping.peGlobal ping & traceroute from 30+ PoPs
Fast.com (Netflix)Quick ISP throughput check

IPAM, DCIM & Documentation

NetBoxIPAM, DCIM, network source of truth
phpIPAMOpen-source IP address management
draw.ioFree network diagramming tool
MX ToolboxMX, SPF, DKIM, blacklist checks
IANA Port NumbersOfficial port & service name registry
RFC EditorFull-text IETF RFC search

Juniper & Cisco Tools

Juniper Apps PortalJTAC, feature explorer, config builder
Juniper TechLibraryFull Junos OS documentation
Cisco DevNetAPIs, SDKs, and automation resources
Cisco Software DownloadIOS/NX-OS/ACI firmware
HPE Aruba DocsHPE Aruba Networking Central technical documentation

AI Infrastructure & Networking

NCCL (GitHub)NVIDIA collective comms library — the AllReduce engine powering GPU training
SONiCOpen-source NOS for AI / cloud fabric switches; used in hyperscale AI clusters
NVIDIA NGC CatalogGPU-optimized containers, pretrained models, and SDKs
MLPerf BenchmarksStandardized AI training and inference benchmark results by hardware
Epoch AI — Compute TrendsAI scaling laws, compute requirements, and training cost analysis
Artificial AnalysisIndependent LLM speed, price, and quality benchmarks across providers
vLLM (GitHub)High-throughput LLM inference engine; supports RoCE/RDMA multi-GPU serving

LLM Tools & Model Hub

Hugging FaceThe model hub: open weights, datasets, and Spaces demos
OllamaRun open-source LLMs locally with one command — Llama, Qwen, Gemma, Mistral
Open WebUISelf-hosted browser UI for Ollama and OpenAI-compatible APIs
LM StudioDesktop app for discovering, downloading, and running local LLMs (GGUF)
Text Generation InferenceHuggingFace production inference server with tensor parallel and quantization
Papers With CodeML research papers with open-source implementations and leaderboards
Arena (formerly LMArena)Crowd-sourced LLM quality leaderboard with blind head-to-head comparisons

WE GET YOU CONNECTED.

Why Interconnected Systems?

For over two decades, Canada's most demanding enterprises have trusted us to design, deploy, and support their critical network infrastructure.

Certified Professionals

Our engineers are certified by Cisco, Juniper, and HPE — among the highest vendor credentials in the industry. You get proven expertise backed by many years of real-world experience, not just paper qualifications.

Networking Experts

From BGP route policy to AI fabric QoS to SD-WAN migrations — we're well-versed in the full technology stack and ready to tackle any challenge you throw at us.

Trusted by Enterprise

Bell, Rogers, BMO, Bruce Power, CTV, and Allstream — Canada's biggest names in telecom, finance, energy, and media have relied on us to keep their networks running.

To learn more about our company, visit our About Us section or contact us directly.

Some of Our Clients

Trusted by Large Enterprise

From national telecom carriers to major financial institutions and broadcast networks — our clients depend on us for their most critical infrastructure.

About Us

Decades of Network Engineering Excellence

Interconnected Systems Inc. has been architecting and delivering mission-critical network infrastructure for Canada's largest enterprises and telecommunications providers for over two decades. Our team brings hands-on expertise across the full technology stack — from physical-layer optical transport to application-aware SD-WAN overlays and AI-driven network automation.

We don't just design networks — we operate them, troubleshoot them, and continuously evolve them as technology and business requirements change. When the biggest names in Canadian media, banking, energy, and telecommunications need complex network problems solved, they call Interconnected Systems.

Get In Touch

Ready to Elevate Your Network?

Tell us about your challenge. Our engineers will respond with a clear, actionable plan.

Toll-Free 1-877-368-1235

Response Time Within 1 business day

Location Canada · Remote Delivery Worldwide

Internetworking Solutions

Design, Implementation, and Support — We've Got It Covered.

Solution Design

Infrastructure Assessment

Implementation & Deployment

Network Automation

AI Fabric & HPC Networking

NOC & Troubleshooting

Cisco

Juniper Networks

HPE & Aruba

Telecommunications

Network Automation

AI Fabric Design

MCP Server — use every tool on this page from Claude or any MCP client

MAC Vendor Lookup

My IP Address

IP Geolocation & ISP Info

Free IPv4 Subnet Calculator

CIDR ↔ Subnet Mask

Wildcard Mask Calculator

IPv6 Subnet Calculator

Network Path Traceroute / MTR

Advanced Path Trace (gtrace)

TCP Port Scanner

Packet Header Size Reference

Connection Quality Tester

CIDR Quick Reference

TCP Throughput Calculator

Real-World TCP Throughput

IP Reputation & Blacklist Check

URL / Domain Threat Check

AI Fabric Reference Library

GPU Fabric Topology Comparator — Rail-Optimized vs Standard

LLM GPU & VRAM Calculator

AI Fabric Planner — RoCEv2 GPU Cluster

AI Power, Cooling & Rack Planner

AI TCO Estimator — On-Premise vs Cloud

Training Data & Storage Sizer

Parallelism Strategy & Scaling Recommender

Inference Serving Simulator

vLLM / Ollama / K8s Config Generator

Cloud Provider Status

ASN Lookup

BGP Prefix Lookup

IP BGP Origin Lookup

Live Route Views Terminal

BGP Looking Glass Directory

DNS Record Lookup

Reverse DNS (PTR) Lookup

DNS Propagation Checker

Email Security (SPF / DMARC / MX)

DNSSEC Validator

IP Blacklist / RBL Checker

Email Address Validator

IPv4 ↔ Binary / Decimal

IPv6 Expander / Compressor

Bandwidth Calculator

IP Range → CIDR Blocks

Mist Toolbox — Chrome & Edge Extension

RF Power Converter

Free Space Path Loss (FSPL)

EIRP Calculator

Link Budget Calculator

WiFi Signal Quality

Wi-Fi Channel Reference

Codec Bandwidth Calculator

VoIP Call Capacity

MOS / Voice Quality (E-Model)

DTMF Tone Generator

SIP Server DNS Lookup

SIP Response Code Reference

High Attention CVEs

RFC Daily Brief

Pricing Data & Sources

Multi-Cloud Interconnect & Cost Planner

Multi-Cloud VPC/VNet/VCN CIDR Planner & Overlap Detector

Egress & Data Transfer Cost Optimizer

Private Endpoint / Service Connect Cost & Sizing

Hybrid Connectivity Options Comparator