IP Transit: The Technical Deep Dive

If you run a network - whether it's an ISP, a hosting provider, or an enterprise with specific connectivity requirements - you've probably heard the term "IP Transit" thrown around. But what actually is it? And more importantly, do you need it?

This post breaks down IP Transit from first principles, explains who benefits from it, and pulls back the curtain on how we architect our transit service at AS152590.

What Is IP Transit?

At its core, IP Transit is a service where you pay a provider to route your traffic to the entire Internet.

When you connect to us via an IP Transit port, we announce your IP prefixes (your public address space) to the global BGP routing table. Your traffic flows through our network, out to our upstream providers, and eventually reaches its destination anywhere on the Internet.

Transit vs. Peering: What's the Difference?

This trips people up constantly, so let's clarify:

Peering is a settlement-free agreement between two networks to exchange traffic destined for each other's customers. If you peer with Google, you can reach Google's services directly - but not the rest of the Internet.

Transit is a paid service where your provider agrees to route traffic to any destination on the Internet, not just their own networks. This is what lets you reach the 100,000+ autonomous systems that make up the global routing table.

The Analogy

Think of peering like having a private road connecting your house to your neighbor's house. Fast, direct, no tolls.

Transit is like paying for access to the entire public road network. You can drive anywhere, but you're paying someone (your ISP) who already has agreements with all the highway operators.

Most production networks use both - peering for high-volume, direct connections to major content providers (Netflix, Google, Cloudflare), and transit for everything else.

Who Actually Needs IP Transit?

Internet Service Providers (ISPs)

If you sell Internet access to end users, you need IP Transit. It's how your customers can reach websites hosted anywhere in the world.

Example Use Case: You run a fiber ISP in regional Australia. Your customers want to access international websites, cloud services, and streaming platforms. You buy IP Transit from us, announce your customer address blocks via BGP, and we handle routing to the global Internet.

Why You'd Choose Us:

• Blended upstreams: We maintain connections to multiple Tier-1 carriers. If one has a routing issue or degraded performance to a specific region, our BGP optimizers shift your traffic to a better path automatically.
• Local peering included: Because we peer at EdgeIX and IX Australia, traffic destined for major Australian content providers takes a direct, local path instead of tromboning through international links.
• Volume pricing: ISPs push serious traffic. We price accordingly with aggressive per-Mbps rates for committed bandwidth.

2. Hosting Providers & Data Centers

If you're running infrastructure that needs to be globally reachable - web hosting, game servers, SaaS platforms - IP Transit is your lifeline.

Example Use Case: You operate a hosting platform with hundreds of customer websites. Those sites need to be reachable from any ISP in any country. You announce your provider-independent (PI) address space via BGP, and we make sure the world knows how to route to you.

Why You'd Choose Us:

• DDoS protection included: Hosting attracts attacks. Our upstream providers scrub volumetric DDoS traffic before it saturates your port. This is included at no extra cost because security shouldn't be an upsell.
• Low-latency paths: We optimize routes via both US-heavy and APAC-optimized upstreams. A user in Singapore gets routed via our Asia-Pacific transit, while US traffic takes the most direct trans-Pacific path.
• BYO address space: You own your /24 (or larger). We'll announce it globally under our AS, but it remains portable if you ever want to multi-home or switch providers.

3. Enterprises with Specific Connectivity Needs

Most businesses can get by with an NBN connection or a generic "business Internet" circuit. But if you're running latency-sensitive applications, need guaranteed uptime, or require control over routing policy, IP Transit might be your answer.

Example Use Case: Your company headquarters runs a private cloud that needs to be accessible from multiple branch offices and remote workers globally. You want dedicated, uncontended bandwidth with the ability to control routing paths during network incidents.

Why You'd Choose Us:

• Uncontended bandwidth: Your 10G port is yours. No oversubscription, no "up to" speeds. You get the full 10 gigabits, all the time.
• BGP community support: You can influence how we route your traffic via BGP communities. Want to avoid a specific upstream during peak hours? Tag your announcements. We'll honor it.
• Direct cloud connectivity: Our peering at major exchanges includes cloud on-ramps. Traffic to AWS, Azure, and Google Cloud takes the shortest possible path.

How Our IP Transit Service Works

Multi-Upstream Architecture

We don't resell a single upstream and call it a day. We maintain diverse transit relationships with multiple providers:

• Tier-1 Global Carrier (US-Heavy Routes): For traffic to North America and Europe
• Regional APAC Carrier: Optimized paths to Southeast Asia, Japan, and China
• Content-Heavy Transit Provider: Direct routes to major CDNs and cloud platforms

This diversity means:

1. No single point of failure: If one carrier has a routing issue, we automatically shift traffic to an alternate path
2. Optimized latency: Traffic to different regions takes the most direct route available
3. Better pricing: Competitive pressure between upstreams keeps our costs (and your rates) lower

BGP Route Optimization

We run custom BGP optimizers that continuously analyze:

• AS path length (shorter is usually better)
• Historical latency data (measured via active probes)
• Communities received from upstreams (some paths are tagged as "backup only")

When multiple paths exist to the same destination, we prefer the route with the best combination of low latency and high reliability. This happens automatically - you don't need to manage it.

Internet Exchange Peering

In addition to paid transit, we maintain 100G peering ports at:

• EdgeIX (NextDC S1)
• IX Australia (Equinix SY1)

These exchanges give us direct, settlement-free peering with hundreds of networks including:

• Google (AS15169)
• Cloudflare (AS13335)
• Amazon AWS (AS16509)
• Akamai (AS20940)
• Major Australian ISPs and content providers

What this means for you: If your users are accessing Netflix, YouTube, or any major CDN, that traffic likely takes a single-hop path through our peering sessions instead of transiting through expensive upstream links. Faster for them, cheaper for you.

Connectivity Options

We offer flexibility in how you connect to our network:

Physical Cross-Connect (Recommended)

Available at:

• NextDC S1 (Macquarie Park)
• Equinix SY1, SY2, SY3, SY4, SY5 (Sydney)

How it works: You order a fiber cross-connect from your rack to ours. We hand you off a clean Layer 1 or Layer 2 circuit. You run BGP directly with our routers.

Why we recommend it:

• Maximum throughput (no virtual fabric overhead)
• Lowest possible latency (direct fiber, no intermediate switches)
• Better troubleshooting (simpler layer 1 topology)

Virtual Cross-Connect (VXC)

Available via:

• Megaport: Instant provisioning from any Megaport-enabled location
• EdgeIX: Direct Layer 2 handoff from EdgeIX fabric
• IX Australia (IAA): Virtual circuit from IAA's SDN platform

How it works: You create a virtual circuit (VXC) through your chosen fabric provider. We accept the connection on our side. You establish a BGP session over the VXC.

Why you'd choose this:

• Fast deployment: Circuits can be lit in minutes via portal provisioning
• Lower commit options: Many customers use VXC for smaller bandwidth tiers (1G, 10G) before scaling to physical
• Geographic flexibility: If you're not physically in our data centers, a VXC can bridge the gap

Performance considerations: Virtual circuits add a small amount of latency (typically less than 1ms) due to the additional switching layer. For most applications, this is negligible. If you're running ultra-low-latency trading systems, go physical.

Technical Specifications

Port Speeds Available

• 1 Gigabit Ethernet: Entry-level transit for smaller networks
• 10 Gigabit Ethernet: Sweet spot for most ISPs and hosting providers
• 100 Gigabit Ethernet: High-volume customers, data center operators

IPv4 and IPv6 Dual-Stack

Every port we provision supports native dual-stack by default:

• You announce your IPv4 prefixes (minimum /24)
• You announce your IPv6 prefixes (minimum /48)
• Both protocols route simultaneously with no extra configuration

This is not optional, and it's not an add-on. IPv6 is the future, and we're not pretending otherwise.

BGP Communities for Traffic Engineering

We offer full BGP community support for customers who want granular control over their routing:

Example: You notice traffic to a specific prefix in the US is taking a suboptimal path via one of our upstreams. You tag your announcement with 152590:2001 to prepend your AS twice when advertising to that specific upstream. Traffic shifts to an alternate provider automatically.

DDoS Protection Included

Volumetric attacks are inevitable if you run a public-facing service. We scrub attack traffic upstream before it saturates your port.

How it works:

1. You detect an attack (or we detect it via our flow monitoring)
2. You announce the attacked prefix with community 152590:666
3. Traffic to that prefix gets routed to our upstream DDoS scrubbing infrastructure
4. Clean traffic is forwarded to your port; attack traffic is dropped

Important: This is not a magic "block all attacks" button. Sophisticated application-layer attacks (e.g., HTTP floods) still require mitigation at your end. But volumetric UDP/SYN floods - which can easily saturate a 10G port - get handled before they reach you.

Who This Service Is NOT For

Let's be honest: IP Transit isn't for everyone.

You probably don't need this if:

• You're a small business with basic Internet needs → Get NBN or a standard business connection
• You don't have your own IP address space → You need provider-assigned IPs first
• You don't have routing expertise → Running BGP incorrectly can black-hole your entire network
• You're looking for "faster Internet for gaming" → This is carrier-grade infrastructure, not consumer service

You might need help, not transit, if:

• You want redundancy but don't know how to configure BGP multihoming
• You're not sure if your existing provider is giving you optimal routes
• You're considering IP Transit but unsure if it's the right move

Call us. Our engineers will tell you honestly whether this service makes sense for your use case, or if you're better off with a different solution.

Pricing Model

We price IP Transit based on committed bandwidth with burstable overages:

• You commit to a baseline data rate (e.g., 1 Gbps)
• We measure your 95th percentile usage over each billing period
• If you stay under your commit, you pay the base rate
• If you burst above, you pay for the additional Mbps used

Volume discounts apply - the more bandwidth you commit, the lower the per-Mbps rate. ISPs moving hundreds of gigabits get significantly better pricing than someone running a single 10G port.

No pricing on this page because it's genuinely custom. A hosted PBX provider with 2 Gbps commit has different requirements than a regional ISP with 50 Gbps. Contact us for a real quote.

Performance Metrics (As of March 2025)

Since launching our AS in 2024, we've maintained:

• 99.95% uptime (measured at BGP session level)
• Sub-15ms average latency to EdgeIX and IX Australia peers
• Zero routing incidents caused by our configuration errors
• Sub-5-minute failover during upstream provider outages (measured during real incidents)

These aren't marketing numbers. They're measured, logged, and reported monthly to transit customers.

What's Next: Geographic Expansion

Our current footprint is Sydney-only. We're actively engineering a Melbourne expansion (NextDC M1 and Equinix ME1) which will launch in 2025.

When it goes live:

• Customers can order dual-homed transit across both cities
• Traffic engineering policies will span metro areas
• VXLAN-based Layer 2 extension lets you run a stretched network across NSW and VIC

Follow our press room for updates.

Summary: Should You Buy IP Transit?

Yes, if:

• You run an ISP and need to route customer traffic globally
• You operate hosting/SaaS infrastructure that must be reachable from anywhere
• You need uncontended, SLA-backed bandwidth with routing control
• You have provider-independent address space and BGP expertise

No, if:

• You're looking for basic "business Internet" (get NBN or a standard circuit)
• You don't have experience running BGP
• Your use case doesn't justify the cost of dedicated infrastructure

Not sure? Call us.

Our NOC engineers will walk through your requirements and tell you honestly whether IP Transit makes sense, or if there's a better solution for your needs.

02 4398 7089 | hello@theitdept.au

Because we'd rather have a happy customer with the right solution than upsell you into something you don't need.