Ethernet Demarcation

In the previous Blog Post I wrote about MEF services and some of the challenges that Carriers and Customers experience.  As I said before MEF has done a fantastic job of defining services and service architectures but sometimes there is an issue in the implementation that cannot be allocated for when designing a service.  This observation is not without anecdotal evidence as mentioned before, I have had a number of conversations with Carriers in the last 6 months and they have confirmed issues with how they implement MEF services.

We do have a very good solution (staring us in the face ) aimed at ethernet service-continuity called, Ethernet Demarcation.  The entire idea is to have a network element that provides a clearly defined area of responsibility, visibility of the service (regardless of who is carrying the traffic) and advanced troubleshooting capabilities to quickly investigate and solve issues in the network when they arise.

Clearly Defined Responsibiility

In order to clearly define roles and responsibilities in a service we need to clearly define the points in the network where responsibilities could (and do) change hands.  This has been done in the TDM world as long as I can remember with NTBA’s (in ISDN) or Fiber Termination Units (with SDH/SONET).  In the age of IP/Ethernet, Packets and “Clouds”, these points aren’t so clearly visible because they are often times logical rather than physical.

By ensuring that the Physical and Logical service demarcation is manifested in the same way (i.e. in the same network element) a clear network “border” is then introduced.  Carriers can then avoid the “finger pointing” issues inherent with blurred lines of responsibility.

Service Visibility and Troubleshooting

As stated earlier many Ethernet services are implemented across various carriers’ networks and service continuity is an issue.  Ethernet has a number of sophisticated tools for analyzing and counting the packets that come through an interface but there is no way of really guaranteeing that a packet gets to it’s endpoint, other protocols are relied upon to do that.  Currently Carriers rely on their installed switches and/or routers to maintain visibility into the service across the network but there are issues with this approach.  Firstly, getting administrative privileges on a foreign carriers’ network elements is, at best, taboo; secondly switches and routers add significant delay and cost; thirdly because of their design switches and routers have limitations to the troubleshooting capabilities (1 way measurements and loopbacks come to mind).

Ethernet Demarcation devices can (and do) assist Carriers by providing much more granular and real-time visibility and troubleshooting capabilities.  Demarcation units sit in the service path at points in the network where responsibilities change hands (Customer to Network/Network to Network) and provide critical visibility metrics to the Service Owner.  Ethernet Demarcation is being provided by a number of Vendors now but they all have a few things in common:

• IP Aware QoS Implementation
• Small Form Factor – Only a few ports are necessary
• Priced like a business CPE – Cost cannot be a barrier

When it comes to implementing a  device that solves the aforementioned issues, these features are a must but a long way from being adequate.  Further features must include:

• Extremely Low Latency
• Advanced Synchronous-like troubleshooting
• Pre-service testing for continuity

If the demarcation device is a switch or a router based architecture then the idea of extremely low latency (meaning 5-15 microseconds through the unit) is, by design, unattainable.  A different fundamental design is needed and a few vendors have begun to implement a “chip on a wire” type of technology.  This has recently become available since programmable chip technologies have reached a point where the price and complexity is sufficient for this type of use.  These chips, called FPGA’s (Field Programmable Gate Arrays), are now available that can be programmed to perform all of the above mentioned features with minimal latency.  Without the “store and forward” architecture that a switch or a router uses one of these units can, theoretically, shove packets through the unit in microseconds.

There are a few Vendors out there who are now producing such a unit.  These units are now being implemented by carriers as CPE, Carrier-Interconnect and sometimes as “Watchtowers” in the network for visibility.  In all of these situations they provide a wide reaching visibility at the service layer for the Service Provider who “owns” the customer.  Furthermore they provide pre-service testing to ensure a service can be provided prior to turning it up as well as advanced troubleshooting capabilities to be able to pinpoint a fault in the network at the service layer regardless of how many carriers or transport technologies may be involved in the service.

Tagged as: Convergence, Demarcation, Ethernet, MEF