Cross-Connects and the Meet-Me Room: The 30 Metres That Decide Your Latency, Cost, and Lock-In

Space and power get negotiated line by line. Connectivity usually gets a shrug and a monthly recurring charge nobody revisits. That is backwards. In a colocation facility, the short run of fibre or copper between your cabinet and the meet-me room is where latency, resilience, recurring cost, and your ability to change providers are all quietly determined.

This is what that path actually looks like, and what to ask before you sign.

What a meet-me room is

The meet-me room (MMR) is the neutral interconnection space inside a carrier-neutral facility. Every carrier, transit provider, cloud on-ramp, exchange, and tenant terminates there. A cross-connect is a physical circuit — usually single-mode fibre, sometimes copper — that runs from your cabinet to a specific port belonging to a specific provider in that room.

Nothing about it is virtual. It is a labelled cable on a patch panel, installed by a facility technician, billed monthly, and documented in a database that is only as accurate as the last person who touched it.

Some larger campuses have more than one MMR, occasionally in separate buildings. That detail matters more than anything else on this page, and we come back to it below.

Why the charge is not the cost

A cross-connect is typically billed as a small installation fee plus a monthly recurring charge per circuit. Individually it is minor. In aggregate it is not, because the number of circuits grows with the architecture, not with the business:

• Two upstream carriers, redundant, is four cross-connects if you terminate on separate routers.
• A cloud on-ramp with a redundant pair is two more.
• An internet exchange port, redundant, is two more.
• Any private circuit to another tenant in the same building is one more per path.

That is eight to ten circuits before anything unusual happens, each one recurring, each one on a contract that outlives whoever ordered it. The cost to watch is not the first circuit. It is the tenth, ordered three years later for a project that ended, still billing.

The diversity question most tenants never ask

Buying two carriers means very little if both cross-connects leave your cabinet through the same overhead tray, enter the same riser, and land in the same MMR on the same patch frame. A single accident in one pathway takes out both. You have paid for redundancy and received two copies of one failure domain.

Real diversity has to be verified at four levels:

1. Cabinet exit — two distinct pathways leaving your cage, ideally opposite corners.
2. Riser and tray — physically separate vertical and horizontal routes through the building.
3. MMR — separate rooms where the campus offers them, or at minimum separate frames.
4. Building entry — separate conduits from the street, ideally from different directions.

Ask the facility for a written path diagram, not a verbal assurance. If they cannot produce one, that itself is the answer.

Latency: small numbers that compound

Light travels roughly 5 microseconds per kilometre in fibre. Inside one building, the distance between your cabinet and the MMR contributes a negligible fraction of a millisecond. So the intra-facility run is rarely where latency is won or lost.

Where it does matter:

• Campus buildings. A cross-connect between two buildings on the same campus adds real distance and, more importantly, adds splice points and patch panels. Each connector adds insertion loss, and every extra hand-off is another thing that can be disturbed.
• Cloud on-ramps. A direct cross-connect to a cloud on-ramp in the same facility removes an entire internet path. For latency-sensitive or bandwidth-heavy workloads, that is the single biggest available improvement.
• Exchange peering. A port on a local internet exchange can shorten the path to regional networks considerably compared with sending the same traffic to a transit provider's nearest hand-off city.

The question to ask is not "how long is the cable." It is "how many networks am I traversing to reach the destination that matters."

Single-mode, multi-mode, and why it keeps costing people money

Multi-mode fibre is cheaper on optics and fine for short runs within a cage. Single-mode is the standard for anything leaving your footprint, and it is what carriers terminate on.

Two failures repeat constantly:

• Ordering a cross-connect and discovering the optics purchased are the wrong mode or wrong wavelength for the provider's hand-off, delaying turn-up by a week.
• Running multi-mode inside the cage, then needing to reach a second-floor cage later and having to re-pull.

Confirm the fibre type, connector type, and expected optic on both ends in writing before the install ticket is raised. This one paragraph saves more turn-up delays than any other.

Lock-in is built here

Migration difficulty is not mostly about servers. It is about circuits. A tenant with twelve cross-connects, several under contracts with different end dates, a few undocumented, and one that nobody can identify but is afraid to unplug, is far harder to move than one with a clean, documented interconnect map.

To keep your options open:

• Keep an interconnect register: circuit ID, provider, both endpoints, port, contract start and end, business owner, and what breaks if it is removed.
• Align contract end dates where possible, so you are not perpetually inside somebody's renewal window.
• Prefer terminating carriers onto your own patch panel inside the cage, so a provider change is a re-patch, not a re-cable.
• Label both ends of every circuit at install, with the circuit ID as written on the invoice.

That last habit is the difference between a two-hour provider swap and a two-day investigation.

What good looks like in the cage

A well-built interconnect layout has a dedicated patch panel for external circuits, physically separated from internal switching. Every port on that panel maps to one row in the register. Fibre is routed on its own pathway, never with power, with bend radius respected and slack coiled in a managed loop rather than looped over a cable arm.

When a carrier engineer arrives to test, they should be able to identify the correct port from the label alone, without a phone call.

Practical questions before signing

• How many meet-me rooms are there, and can I get true room-diverse paths?
• Who owns the MMR — the facility, or a third party who bills separately?
• What is the standard installation lead time for a cross-connect, and the expedite option?
• Are there restrictions on which carriers I can reach, or preferential pricing arrangements?
• Can I bring my own dark fibre into the building, and what does that entry cost?
• What happens to my circuits if I add a second cage later — are intra-facility circuits billed the same way?

The short version

Cross-connects are the smallest line item with the largest strategic consequence in a colocation agreement. Price them as an architecture decision, verify diversity physically rather than contractually, and keep a register from day one.

StackTrue installs, labels, tests, and documents cross-connects across Canadian facilities, and hands back a full interconnect register with every deployment. If you inherited a cage with circuits nobody can identify, an audit is usually a one-day job.