100119

Carrier Ethernet, and how it’s different

ADVA Optical Networking recently announced that it has joined the Ethernet Alliance, with yours truly acting as chair of the Carrier Ethernet Subcommittee. Since that time I have received dozens of emails, most of which have asked two fundamental questions which I hope to answer here in this post.

The first question is the somewhat easier of the two: “What is the Ethernet Alliance?” While the Ethernet Alliance (EA) is well known and established in the Enterprise space, it is somewhat newer to the Carrier space. You can think of the EA as a place for people with similar interests and their affiliated companies to go and work on anything and everything Ethernet, building consensus in the process. Unlike most standards bodies, the EA encourages company participation, affiliation, and positioning. One of the many examples of EA success over the years has been the work of the High Speed Ethernet Subcommittee in helping build consensus on the definition of 40/100GE. The EA has had similar success in the areas of Data Center, 10GBASE-T, and Energy Efficient Ethernet, to name just a few, and it has now turned its attention to one of the hottest areas of growth in Ethernet, the Carrier space.

Which brings us to the second question: “What exactly is “Carrier Ethernet” and how is it different than plain vanilla “Ethernet”?

While the term “Carrier Ethernet” has evolved over the years, I would define it a bit generically as any Ethernet product or service that meets the demands of Carriers, with the recognition that Carriers needs are constantly changing. Basically, Carrier Ethernet appends standard Enterprise Ethernet with additional features and protocols that provide the capabilities Carriers need to deliver services to their customers.

In a recent posting here at Technically Speaking, John Messenger discusses some of the next-generation Ethernet encapsulation methods that Carriers will need/use to provide Ethernet services to their customers. To re-use one of his examples, in traditional Ethernet each end-user has their own MAC address, which works great for LAN’s (essentially isolated islands of connectivity). However, when Carriers start interconnecting disparate LAN’s, transporting millions of user’s data traffic in the process, there is a need for MAC address isolation. While in an ideal world, everyone has a unique MAC address, unfortunately that is not always the case as sometimes NIC’s and the MAC addresses embedded in them have been cloned. Ethernet’s traditional flat addressing scheme can wreak havoc when two independent LAN’s have been interconnected by a Carrier. IEEE standard 802.1ah, also known as MAC-in-MAC, solves this problem by adding hierarchical MAC addressing to Ethernet.

Another example would be VLAN tagging. Traditional Ethernet uses a 12-bit VLAN field, which yields 4095 VLAN tags, which is more than enough for traditional Enterprise LAN’s. However, Carriers use VLAN tags to differentiate between customers and services, and with the onslaught of triple-play, HDTV, VoD, etc, 4095 is simply not enough tags. Carrier Ethernet networks support double-tagging, also known as Q-in-Q, where the packets contain both an outer and inner VLAN identifier, thus vastly increasing the number of VLAN tags supported, as specified in IEEE standard 802.1ad.

However, Carrier Ethernet does not necessarily refer to new packet formats alone, as the algorithm used by the switching engines may also be specified, as is the case with Hierarchical QoS (H-QoS). Simply marking packets with priority is not enough, if they still end up stuck in line behind a jumbo packet, waiting on their turn in the scheduler. With H-QoS, bandwidth management techniques are used to make intelligent decisions regarding which is the next packet to get scheduler bandwidth. An analogy might be letting someone behind you in line at a supermarket skip in front of you if they only have one item and you are pushing a full cart. H-QoS is proving indispensible in Carrier networks, allowing them to maintain the expected level of service on latency sensitive traffic when faced with a deluge of non-critical traffic. One example would be separating the voice from data traffic when a consumer is simultaneously using the phone and browser features of their smartphone.

These are just a few examples of how Carrier Ethernet differs from traditional Ethernet. If this type of work seems interesting to you, and you would like to help shape the future of Carrier Ethernet, consider joining the Ethernet Alliance and specifically the Carrier Ethernet Subcommittee. If your company is already a member of the Ethernet Alliance, you can simply go online and sign-up for the group. If your company has not yet joined, then you can still go online, fill out an application, and join the Carrier Ethernet Subcommittee meetings as my guest while your company considers the benefits of membership. Either way, come join the fun!