Server nodes and routes overview

IEPL: an end-to-end dedicated channel

IEPL (International Ethernet Private Line) is a point-to-point Layer 2 line from a carrier. Its defining feature is an exclusive path: traffic from entry to exit never touches public internet backbone routers and never shares bandwidth with other users, so jitter and packet loss stay within a very narrow range.

That property matters most for two kinds of workloads. The first is long-lived connections — streaming output from AI tools, SSH sessions, database sync. Here a single lost packet can force a retransmission or even drop the whole session, and jitter on a public link turns into visible stalling. The second is real-time traffic — video meetings, live streaming, cloud gaming — where stable latency matters more than raw latency.

The trade-off is cost. Private lines are billed by bandwidth, and the price per Mbps is far higher than ordinary broadband, so IEPL routes usually cover only a few core cities rather than every city. VPNDM concentrates IEPL on the hubs of Hong Kong, Tokyo, Singapore, Los Angeles, London and Frankfurt, because they are also the main interchange points for intercontinental traffic.

One easy point of confusion: IEPL does not mean faster. If the server you are reaching is already in a nearby region, a private line adds little; its value shows up on intercontinental links and during peak hours.

Relay: a nearby entry forwards to the exit

A relay route has two legs: you connect to an entry node near you, and that entry node forwards traffic to the target exit node. Its point is shorter cross-border detours. On a direct intercontinental route, packets may detour through a third country and back; a relay hands that leg to an optimized internal link, so your side only connects to the nearby entry.

For example, reaching South America from Asia on a direct path often goes through the North American west coast and then south — many hops and many autonomous systems. A relay connects you to an Asian entry first, then takes a fixed trans-Pacific link to North America and heads south from there. On your side it looks like you connected to a nearby node, while the actual exit is still in South America.

Relay has the widest range of uses: everyday browsing, streaming, cross-continent work. It is steadier than direct and cheaper than IEPL, and it is the most common type in the route list. One thing to note: relay performance depends heavily on the load of the entry node and the quality of the exit link, so two relay routes in the same city can feel different depending on the entry.

Direct: shortest path, simplest structure

Direct means connecting straight to the exit node with no extra forwarding in between. Its advantage is transparency: you know which data center you are on and where traffic enters and leaves, with no middle layer. For cases that need a clear end-to-end path — testing how a service looks in a specific region, or reaching regional content that is only open to certain areas — direct is the most straightforward choice.

Direct falls short on intercontinental routes. When the physical distance is long and traffic has to use the public internet, the path is decided by peering between carriers, and peak hours bring detours and congestion. So direct routes suit two cases: nearby access, where the target service is not far from you geographically, and targets with a clear exit-region requirement that need no extra path optimization.

Direct is the cheapest of the three, which is why it covers the most cities. If you need a local service in a specific country or region, look for that country's direct routes first; if the experience is not good enough, try a relay or private line in the same region.