Ethernet Private Line and WAN, Measured Where You Need It
Carrier Ethernet links your offices, plants and data centers at Layer 2, point to point or many to many, at a bandwidth the provider commits to carry. TREU finds which providers are on-net at each site and which would hand you to a local carrier, negotiates with your legal team, and stays with each agreement through every renewal.
Book a conversationWhat Ethernet Private Line Is, and Where It Fits
Carrier Ethernet services follow definitions published by MEF (now Mplify), the industry standards body. A private line joins two sites, and an E-LAN joins many as if they shared one LAN. You buy a committed rate (CIR) on a port, and the provider's performance objectives apply to traffic within it. Where a site is off the provider's own network, it buys the last stretch from a local carrier and delivers both as one service.
When it makes sense
- Two sites that need a private, transparent link, with your team running its own VLANs end to end. That is EPL.
- A data center or headquarters reaching many branches over one larger port. That is EVPL, the common hub-and-spoke design.
- Many sites that need to reach one another at Layer 2 for shared servers and storage. That is E-LAN.
- Traffic that needs committed bandwidth, with delay, delay variation and frame loss objectives written into the SLA.
When it does not
In our judgment, Ethernet may be more than you need when:
- Sites mostly reach the internet and cloud applications and can live with internet performance. SD-WAN over internet access may fit better.
- You want the provider to route between sites. Ethernet leaves the routing to your team, so MPLS or SD-WAN may suit you better.
- Two sites need 10 gigabits or more of full-rate, fully transparent capacity on a fixed path. A wavelength often serves them better.
Some of your sites may need something other than Ethernet, and we will name them.
EPL, EVPL or E-LAN
| EPL | EVPL | E-LAN | |
|---|---|---|---|
| Shape | Point to point | Point to point, several on one port | Many sites to many sites |
| How traffic is sorted | Everything on the port goes to one connection | Your VLAN ID picks the connection | Every site joins one shared LAN |
| VLANs and the provider | Your VLAN IDs pass through; no coordination needed | VLAN IDs agreed with the provider | No coordination on the port-based version |
| Best fit | Two sites, with your team running the VLANs | A hub reaching many branches | Many sites sharing servers and storage |
Technical Considerations
Two Ethernet quotes at the same speed can perform very differently. The difference sits in these details.
Committed rate and burst
The provider's objectives apply only to traffic within your committed rate and committed burst. Traffic above them is carried without objectives up to any excess rate you buy, and dropped beyond it. Shape traffic on your router to the CIR for each class, and check the burst size the provider allows at your port speed.
Classes of service
Providers sell a few classes, such as basic, priority and premium, and you set a CIR for each. Priority markings have to be agreed, and one provider bars customers from remarking traffic without approval. The classes on offer can differ between on-net and off-net sites.
Where performance is measured
Some SLAs measure delay and jitter between the provider's own core sites, not between yours. A core-only figure says nothing about the access circuits at each end. Targets also change by distance tier, so a metro figure does not apply to a continental link, and some connections fall outside the latency SLA altogether.
Off-net tails
Where a site is off the provider's network, a local carrier delivers the last stretch through an interconnection between the two. Class options, CIR steps, frame size and upgrades then depend on what that carrier supports.
Frame size
Maximum frame size can differ by access type. Jumbo frames may apply only if the order says so and, off-net, only if the local carrier supports them. Oversized frames are dropped, which matters for storage replication and backup.
Control protocols
Even a port-based private line may not carry every Layer 2 control protocol frame, and EVPL filters more of them than EPL. If your design depends on one, confirm how it is handled before you order.
Equipment and access
The provider usually places a device at each site, which needs space, power and a clear path from the building entry, and battery backup may be yours to supply. The access type can limit what is available: one provider will not multiplex EVPL over its coax network, and at least one carrier is retiring copper Ethernet.
Turn-up and billing
Billing can start before you have tested, as the network family page explains. Test at turn-up, ask for the activation test results, and keep the old circuit live until the new one is accepted.
Why IT and Finance Both Care
For IT and network leaders: a network that performs where your traffic runs
- The right service type at each site, chosen for how traffic actually flows.
- SLAs measured between your sites, not only in the provider's core.
- Off-net sites known before you sign, along with their limits on classes, frame size and upgrades.
- Install dates, turn-up tests and renewal dates at every site chased by TREU, carrier by carrier, instead of by your team.
For finance: fewer surprises after signature
- Exit costs for on-net and off-net circuits compared before you commit. In at least one provider's terms, leaving an off-net circuit early costs every remaining month.
- Moves and upgrades agreed in advance, so a new site or a faster port does not trigger termination charges.
- Billing tied to acceptance wherever the provider will agree to it.
Four Stages of an Ethernet Purchase
See how we work-
Supplier identification
We check which providers are on-net at each of your sites and which would deliver over a local carrier, wherever your sites are. That includes regional, utility and municipal networks.
-
Buying cycle
We match the service type to each site, then compare CIR and classes, where performance is measured, off-net terms and exit costs side by side. We negotiate the commercial terms alongside your counsel.
-
Implementation oversight and accountability
We hold the provider to its install dates and to passing turn-up tests before billing starts, and check the first invoice against the contract.
-
Client and supplier lifecycle management
We track terms and renewal dates, join business reviews, and source new sites and upgrades as you need them. In our experience, the work starts months before the notice date, not weeks.
What to Watch For in an Ethernet Contract
Ethernet contracts look alike until a site moves or a circuit fails. These are the terms that tell them apart.
-
Availability credits.
Check when the outage clock starts, whether separate outages add up, the cap on credits and the deadline to claim. Credits are often the sole remedy.
-
Performance credits.
Latency, jitter and frame loss credits often depend on you noticing the problem, claiming within a set window and then giving the provider time to fix it.
-
What the SLA covers.
Some availability promises include the local loop, and some cover only part of the path. Read which one you are buying.
-
Install dates.
Some contracts commit only to reasonable efforts toward an estimated date, and a miss is not a breach. If the date matters, negotiate a remedy.
-
Early termination.
Leaving an on-net circuit early may cost a declining share of the remaining charges. Leaving an off-net one can cost every remaining month plus the local carrier's charges.
-
Moves and upgrades.
Agree up front that you can move a circuit or raise its speed without termination charges, and find out what the local carrier allows off-net.
-
Changes to the local carrier.
At least one seller can swap the local carrier or move a site onto their own network at their discretion. Ask for notice, and for any diversity you paid for to be protected in writing.
-
The term on each order.
If an order does not state a term, the contract may set one for you. Write the term into every order.
Treat this list as a starting point. TREU is not a law firm. Your counsel gives the legal advice.
Questions Buyers Ask About Ethernet
What is the difference between EPL and EVPL?
EPL is port-based. Everything on the port goes to one connection, and your VLAN IDs and priority bits pass through unchanged, so you don't need to coordinate VLANs with the provider. EVPL is VLAN-based. Your VLAN ID decides which connection a frame joins, so one port can carry several connections, which makes EVPL the usual choice for a hub that reaches many branches.
What is E-LAN, and when would we use it?
E-LAN connects many sites so they behave as if they share one LAN, with the same access to servers and storage. On the port-based version, you can set up VLANs across sites without involving the provider. A related service, E-Tree, lets branch sites reach a central site but not each other.
What is CIR, and what happens to traffic above it?
The committed information rate is the bandwidth the provider commits to carry within its performance objectives. Traffic above it is carried without objectives up to any excess rate you buy, and dropped beyond it. Shaping traffic to the CIR on your own router keeps bursts from being discarded.
What does MEF 3.0 certified mean?
MEF, the industry body that defines carrier Ethernet services, certifies services one at a time. The provider configures a specific service, an authorized test partner tests it against MEF requirements, and the service is listed in MEF's certification registry. Certification covers that named service, not every product or off-net tail the provider sells. Ask which service is certified, check the registry and request the test report.
What is a Type 2 circuit, and does it matter?
A Type 2 circuit reaches a site off the seller's network over a local carrier's access, joined to the seller's network at an interconnection. You still buy one service, but the local carrier's limits on classes, bandwidth steps, frame size and upgrades apply, and in at least one provider's terms it costs more to leave early. We identify which of your sites would be Type 2 before you choose.
What do Ethernet SLAs cover?
Most set objectives for frame delay, delay variation (jitter), frame loss and availability, by class of service and distance. Read where they are measured: some cover only the provider's core between its own sites, not the circuits into yours. Credits are usually capped, must be claimed on time and are the sole remedy.
How long does an Ethernet circuit take to install?
It depends mostly on whether the provider already has facilities at your site. Where it must build, expect the interval to stretch. Some contracts treat the install date as an estimate, so if the date matters, negotiate a remedy. Your own site readiness counts too: space, power and a path from the building entry for the provider's equipment.
Ethernet, MPLS or SD-WAN?
In our judgment, Ethernet fits when you want private Layer 2 links and your team runs the routing. MPLS fits when you want the provider to route between sites with classes of service. SD-WAN over internet access fits sites whose traffic mostly goes to the internet and cloud. We match the service to each site.