Remote IXP vs physical presence

Remote IXP is a model of internet exchange access in which a network reaches a port on the exchange switch through a link provided by a third party (a reseller, transport carrier or EoMPLS service) instead of installing its own router at the IXP site. The alternative is physical presence: colocating your own hardware in the facility that houses the exchange fabric. Both approaches connect to the same peering table. They differ in cost structure, geographic reach and operational burden.

Cost: CapEx of physical presence versus OpEx of a remote port

Physical presence involves capital expenditure: buying an edge router or switch, rack space (colocation), power, optical modules and cross-connects into the IXP fabric. On top of that sits transport cost when the IXP building does not coincide with a node of your own network. The remote model converts those into predictable OpEx: a monthly exchange port fee plus a transport charge to reach that port. For a small network with moderate peering traffic, a monthly remote port is often cheaper than depreciating hardware and maintaining colocation.

• Physical presence: edge router, colocation rack, A/B power, cross-connect, optional transport.
• Remote IXP: a single logical port on the exchange switch plus a transport link from a reseller or carrier.
• CapEx jumps in steps each time you enter a new IXP, while OpEx scales linearly with the number of ports.

Reach, latency and the 10:1 ratio

Remote access opens exchanges your network does not physically reach, which matters when building peering in another city or country. The price is the extra latency added by transport: traffic peered over a remote port travels to the distant IXP and back. A practical rule of thumb here is the 10:1 ratio. It says remote access is worthwhile as long as the traffic volume you can move off transit through peering is clearly larger (on the order of ten times) than the cost and overhead of transport to that IXP. Once the ratio drops, it is cheaper and simpler to buy transit or establish local presence.

With physical presence the latency to the fabric is minimal, because the router sits in the same building, and the longer path only appears on the backhaul to the rest of your network. That is why networks with large, latency-sensitive traffic (CDNs, gaming, live video) usually aim for physical presence in key locations.

Operational burden and when remote IXP makes sense

• Remote IXP removes the duty of on-site hardware service, optics replacement and smart-hands visits.
• A smaller ISP gains access to a large peering table without building a point of presence from scratch.
• Physical presence gives full control over hardware, port redundancy and future fabric expansion.
• The remote model adds dependence on the transport carrier as an extra point of failure on the path to the IXP.

For a smaller operator a sensible strategy is to start remote, validate the real peering volume and return from a given IXP, then migrate to physical presence where traffic and the 10:1 rule justify it. That way investment in a router and colocation only lands in locations that genuinely reduce transit cost and improve path quality.