IEPL private lines
The cross-border hop is the weak link. Stop leaving it to the public internet
Same two endpoints, two very different paths. The public route starts at 200ms and sheds up to 15% of packets in peak hours. An IEPL circuit runs point-to-point at layer 2 under 50ms with 0.01% loss. The difference is not bandwidth — it is whether anyone else is on the road.
Public internet
- Latency
- 200ms+
- Packet loss
- 15%
- Jitter
- Very high
IEPL line
- Latency
- <50ms
- Packet loss
- 0.01%
- Jitter
- Microseconds
What goes wrong on public transit
Cross-border traffic crosses several autonomous systems on a path you do not choose, and that path is at its worst in exactly the hours you need it most.
- Detours: a packet from Shenzhen to Los Angeles may transit Tokyo or Frankfurt first — same distance, double the latency
- Peak congestion: international gateways saturate in the evening, loss climbs past 15%, and retransmits push latency higher again
- Uncontrolled jitter: pushes and real-time sync care about jitter far more than throughput, and jitter is always the first thing to go
What an IEPL circuit actually changes
The carrier reserves a layer-2 circuit between two facilities. It never enters the public internet, the bandwidth is yours, the path is fixed, and the quality can be committed instead of estimated.
- Point-to-point layer 2, so your traffic never shares a queue with the public internet
- Dedicated, committed bandwidth — bursts are not shaped and throughput holds at the rate you bought
- End-to-end monitoring with live latency, loss and jitter, so a fault is isolated to a segment rather than argued about
How provisioning works
Lines are delivered by port specification and committed bandwidth, so both endpoints and the traffic type have to be settled first. Turn-up is typically one to three business days.
- Tell us the workload — RTMP push, cross-border office, data replication, real-time interaction — and the region at each end
- Size committed bandwidth and port count against your peak rather than your average
- An engineer configures the circuit and routing policy, then hands back the endpoint and credentials
- Once live, latency, loss and jitter are visible in real time, with alerting when the link degrades
Engineering detail
Who runs on IEPL
Workloads where stability outranks unit price.
Overseas live streaming
A long RTMP push fails on jitter, not on bandwidth. Fix the path and the encoder stops stepping the bitrate down.
Cross-border office and internal apps
Branch access to headquarters systems, ERP and finance tools behaves very differently on a reserved circuit than on a VPN over public transit.
Real-time data replication
Primary-replica sync and order write-back break the timing budget long before they break the bandwidth budget when retransmits pile up.
Cross-border video conferencing
Meetings that stutter between countries are losing packets, not running out of throughput. More bandwidth will not fix it; a different path will.
Games and interactive sessions
For frame-synchronised work, a steady 60ms is worth more than a 40ms average that spikes without warning.
FAQ
FAQ
How is IEPL different from a VPN or SD-WAN?
A VPN wraps traffic in encryption but still rides the public path, so congestion reaches it exactly the same way. SD-WAN picks intelligently between links, which helps — but the links it picks between are still public. IEPL is a carrier-reserved layer-2 circuit between two sites that never enters the public internet, which is why latency and loss can be committed rather than estimated.
IEPL or MPLS — which should I buy?
IEPL is point-to-point at layer 2: effectively a very long patch cable between two facilities, simple to configure and very stable in delay. MPLS is a layer-3 any-to-any network for connecting many branches with routing policy — more flexible, more expensive, more to operate. For one live push or one replication link between two places, IEPL is the better fit.
Will a private line make my downloads faster?
It fixes latency, loss and jitter, not raw bandwidth. If cross-border quality is bad enough that TCP is constantly retransmitting, the improvement is large and immediate. If the ceiling is the far end rate-limiting you or your own uplink being full, a private line will not move it. We work out where the bottleneck sits before quoting, and say so if a line is the wrong answer.
How is IEPL priced, and what is the minimum?
Pricing follows committed bandwidth and port count, billed monthly, and depends on which two regions the circuit lands in. Cross-border lines are delivered case by case — email support@zapip.net with the workload and both endpoints and you will get a concrete design and a quote.
Get one request working. Scale after that.
Test before you commit, with an engineer on your integration.