You don't own your network, and that's your real lock-in.
You can leave a host in an afternoon. Leaving their network, their IP space, their egress bill, and their upstream's rules is the real lock-in. Here is the one question that exposes it.
In this article 5 sections
- Many hosts do not own their network. They rent transit and resell it, and every promise they make is capped by a contract they do not control.
- We run our own global autonomous system, AS200477, across our datacenters in Athens, Amsterdam, and Toronto: our own BGP, multi-homed on multiple upstreams, peering at GR-IX and NetIX, IPv4 space held directly from RIPE.
- Every claim in this post is on the public record: look AS200477 up yourself on bgp.he.net.
- The forwardable test: five questions to ask any host about their network. The answers separate operators from tenants.
Moving your servers to a new host is a project. Moving your network off a host is a different animal: new IP space to warm up, mail reputation to rebuild, DNS to repoint everywhere, firewall allowlists at every partner who ever pinned your address. That lock-in never shows up as a line item, and it is why the network underneath your host matters more than the badge on their homepage.
The rented backbone
Here is the part of the hosting industry that does not make it into sales decks. Many hosts are tenants. They buy transit from an upstream carrier, rack servers behind it, and resell the connectivity as if it were theirs. The routing policy belongs to the upstream. The maintenance windows belong to the upstream.
When that upstream has a bad day, reprices, or changes policy, your host opens a ticket and waits in the same queue as everyone else. So do you, one level removed, with no visibility into any of it. The tell is simple, it is public, and it fits in one email.
Ask who owns the autonomous system your traffic rides. What comes back?
You're on an operator
They set routing policy, choose the carriers, and fix the network themselves. The SLA is theirs to keep, not to pass through.
You're a subtenant
Routing policy, maintenance windows, and the repair queue all belong to a carrier you have no contract with.
What owning the AS buys you
Our network announces itself globally to the internet as AS200477 across our datacenters in Athens, Amsterdam, and Toronto, registered to us at RIPE. That one fact carries the whole argument, because everything an AS owner controls, we control: which carriers we buy transit from, how traffic enters and leaves, what gets rerouted when something degrades, and how fast any of that changes. We are multi-homed on multiple upstreams, including global Tier-1 transit and regional incumbents, and we peer at GR-IX and NetIX. No single provider is our ceiling, which means no single provider is yours.
The addresses are ours too. Your services ride IPv4 space held directly from RIPE, assigned to us, not rented from a carrier who can reclaim them. Your IP reputation, the thing your deliverability and your partners' allowlists actually depend on, lives in address space that does not change hands when you scale or when we renegotiate transit.
Our own BGP, announced globally across Athens, Amsterdam, and Toronto as AS200477
Multi-homed: global Tier-1 transit and regional carriers
Peering at GR-IX and NetIX internet exchanges
IPv4 space held directly from RIPE
The global routing table is the one place a host cannot do marketing.
The egress trap, answered
The hyperscaler egress fee is a lock-in dressed as a line item: you pay to move your own data out, which puts a price on every future exit before you have planned one. We have written about the mechanics in the exit tax, so here is only the networking half of the answer. A host that resells transit by the gigabyte has to meter you, because their supplier meters them. A host that owns the pipes does not. There is no per-gigabyte line on our network because we buy transit on committed capacity from several carriers and peer at exchanges, so there is no per-gigabyte meter to pass on to you.
The doctrine at the core
The network is built the way the storage is built, and for the same reason. Cisco ASR 9000 at the Athens edge and Juniper MX204 in Amsterdam, feeding a Cisco Nexus 9000 EVPN/VXLAN fabric: a small, standardized set of platforms, run by the same engineers who answer your ticket. The hardware doctrine applies edge to spine: we deviate only when the job earns it.
Five questions to ask any host about their network
This list is designed to be forwarded to whoever you are evaluating, including us. The answers separate network operators from network tenants in one email.
Do you own your AS?
Ours: yes. AS200477 across Athens, Amsterdam, and Toronto, registered to us at RIPE, verifiable in the public routing table.
How many upstream carriers do you have?
Ours: multiple, including global Tier-1 transit and regional incumbents. Losing one carrier is a reroute you never notice.
Do you peer at an exchange?
Ours: yes, at GR-IX and NetIX. Local and regional traffic takes the short path instead of tromboning through a carrier.
Who changes your routing when it needs changing?
Ours: we do, on our own edge, without a change request to anyone upstream.
Is egress metered?
Ours: no per-gigabyte line, because we buy transit on committed capacity and peer at exchanges. You pay for the machine, not the meter.
Size a build on a network we actually own.
Seven questions, a real spec, a price range. The network underneath is the one described above.
Open the Sizing DeskStraight answers
Can I verify any of this myself?+
Yes, and you should. AS200477 is in the global routing table: bgp.he.net or any looking-glass will show you the AS, the upstreams, the exchange peering, and the address space. If a host's network claims cannot be checked from the outside, treat them as copy, not fact.
What happens if one of your carriers fails?+
Traffic reroutes over the remaining upstreams and the exchange.
Do my IP addresses change if I grow?+
No. The space is assigned to us by RIPE, not leased from a carrier, so scaling up inside our network does not mean renumbering. Your addresses and the reputation attached to them stay put.
Does who owns the network affect my email deliverability?+
Yes: your sending reputation lives on IP space. Ours is held directly from RIPE, so it does not change hands, and the full argument is in why your email lands in spam.
If your host's network belongs to someone else, so does your exit. Ask the five questions, check the routing table, and when you want the answers walked through on a whiteboard, come meet the people who run the AS.
You’ve reached the end.
Put the reading to work
What are
you running?
Tell us about the workload, traffic and recovery requirements. Start with the server builder, or talk through the details with an engineer.
Written by the engineers who build and run the infrastructure it describes.Talk to an engineer
