Carrier Ethernet is a family of standardized services that telecom carriers use to deliver Ethernet connections between business locations, data centers and cloud on-ramps across their own networks. You plug into an Ethernet port at each site, and the carrier carries your traffic to the other end as if the sites were joined by a very long network cable or a shared switch. The service types and their attributes are defined by MEF, the industry body formerly called the Metro Ethernet Forum, which gives buyers a common vocabulary for comparing quotes.
At a glance
- Carrier Ethernet is a private Layer 2 service: it links your sites to each other, or to a data center or cloud on-ramp, not to the internet.
- The main service types are E-Line (point-to-point), E-LAN (multipoint) and E-Tree (hub-and-spoke), with port-based and VLAN-based versions of each.
- Each service has a bandwidth profile, often with a committed information rate, and the contract may include service levels for availability, delay, jitter and frame loss.
- Most services are delivered over fiber; some use copper, such as Ethernet over copper, or wireless access.
- It is not encrypted by default, and service levels, reach and price vary by carrier and by address.
What problem it solves
Companies with more than one location need those locations to share applications, file servers, phone systems and backups. Sending that traffic across the public internet works for many purposes, but performance varies and no single provider controls the whole path. Older private options such as T1s and legacy leased lines were slow and expensive per megabit.
Carrier Ethernet gives you a private connection with a defined amount of bandwidth that uses the same Ethernet technology your internal network already runs. Your team can treat a remote office or data center much like another part of the local network, and the carrier commits in the contract to the capacity and, depending on the contract, to performance targets.
How it works
User network interface (UNI). The carrier installs a port at each site, usually on a small device in your telecom room. That port is the demarcation point between your equipment and the carrier’s service, and it is where service levels are usually measured.
Ethernet virtual connection (EVC). The carrier builds a logical connection between UNIs across its network. Point-to-point services are E-Line, sold as an Ethernet private line (EPL) that uses the whole port or an Ethernet virtual private line (EVPL) that lets several connections share one port by VLAN. Multipoint services are E-LAN. E-Tree lets a hub talk to every branch but keeps branches from talking to each other. Carriers also sell wholesale Ethernet access to other providers, which is how many off-net locations are reached.
Bandwidth profile and classes of service. Each EVC has a committed rate, sometimes with an excess or burst rate above it that is delivered only when capacity allows. Some providers offer several classes of service so that voice can be treated differently from bulk data.
Monitoring. Ethernet operations, administration and maintenance (OAM) tools let the carrier, and sometimes the customer, check that the connection is up and measure delay, jitter and loss against the contract.
Inside the carrier’s network the traffic may ride on MPLS, optical transport or switched Ethernet. That is invisible to you; what you buy is the Ethernet service at each end.
When it matters for buyers
- Linking a headquarters to a data center or colocation site where steady, high-volume traffic such as backups and replication would be unpredictable over the internet.
- Building a private underlay for SD-WAN. Many organizations pair a Carrier Ethernet circuit with internet links under an SD-WAN overlay.
- Reaching cloud providers privately, by running a circuit to a data center that hosts a cloud on-ramp or buying a carrier’s cloud connection product.
- Replacing legacy circuits such as T1s that are being retired or are too slow.
- Planning a new site. Availability depends on whether the carrier’s fiber reaches the building; off-net addresses can mean construction charges and long lead times.
Questions to ask vendors
- Is the address on your network, and if not, what construction, cost and lead time are involved?
- Which service type are you quoting (EPL, EVPL, E-LAN), and what VLAN and control-protocol behavior does it support?
- What is the committed rate, and is any burst above it guaranteed or best-effort?
- What does the service level agreement cover (availability, delay, jitter, frame loss, time to repair), and what credits apply if you miss it?
- Is the circuit protected with a diverse path, and will you document the physical route?
- What maximum frame size (MTU) is supported end to end?
- Can I see performance data, and how quickly can bandwidth be changed?
See our overview of private networking for how Carrier Ethernet fits alongside other private WAN options.
How it differs from MPLS
Both are private WAN services, but they work at different layers. Carrier Ethernet is a Layer 2 service: the carrier moves Ethernet frames and you run your own routing between sites. An MPLS IP VPN, the most common form of MPLS sold to businesses, is a Layer 3 service: the carrier routes your traffic between sites and usually offers several classes of service. Carrier Ethernet gives you more control and is often simple for a few sites in a region; MPLS VPNs can be easier when you have many sites and want the carrier to handle routing. Pricing depends on the carriers, locations and bandwidth involved, so compare quotes for the same sites rather than assuming one is cheaper. Carriers often use MPLS inside their networks to deliver Carrier Ethernet, which adds to the confusion.