The Mobile World Congress was full of ideas about apps, lifestyle, mobile workflows and processes that can be enhanced by mobility. I was stunned and amazed at how little promotion there was about Ethernet and Mobile Backhaul.
In 2009 the lack of adequate Mobile Backhaul implementations were all over the press in the US and UK. One would think that the participants at the MWC would be promoting their latest mobile broadband solutions with fanfare. It seems that the Elephant in the room will remain uncomfortably ignored.
This should be (is) the paramount issue in Mobile Networking today and equipment vendors should be investing resources to come up with solutions. All of the entertainment, applications and business solutions promoting data usage are useless if the bandwidth end-to-end is not there to support them.
Fortunately I was in a position to engage multiple Mobile Carriers and Telecommunications Vendors at the MWC. Obviously the paramount concern around bandwidth has been answered with Ethernet in the backhaul but the devil is in the details. Throwing bandwidth at the problem has proven time and time again to be a only a band-aid.
From the discussions I had with operators and vendors there were a few common themes and/or concerns. Here they are in summary.
The single most important point Mobile Carriers are concerned about today is ensuring visibility into the service layers of their network. In other words, they want to measure, monitor, trend and manage their networks as accurately as possible. Currently TDM/ATM services are used for backhauling and there are tools inherent to these services which provide a level of visibility and comfort to Carriers. With the move to Ethernet in the backhaul, mobile carriers are loathe to give up these capabilities.
In many cases these carriers do not own their own wireline backhaul infrastructure. These carriers could just trust their leased infrastructure to carry the traffic according to the SLAs they have contracted, however we know in this day of “elastic bandwidth” via IP/MPLS that SLA’s are easily compromised.
I can summarize this with a quote from one carrier I spoke with at the show who said,
I ordered a 20 Mbit ethernet service with a maximum latency of 10 milliseconds in October 2009 all under SLA. I measured it last week and I was getting 7Mbit with a latency of over 40 milliseconds. What I thought was a dedicated service was actually an MPLS based service with no guarantees, so we invoked the SLA penalties.
Active Service Monitoring
Standards based Ethernet OAM is the answer most vendors and operators have for ensuring active continuous monitoring of the network. These technologies can be implemented in a way that the operator would know of any issues in the network, most likely, before a customer notices any problems. Such an implementation is necessary when actively monitoring a mobile network.
What I thought was a dedicated service was actually an MPLS based service with no guarantees, so we invoked the SLA penalties
However, the devil is in the details. Ethernet based OAM requires a lot of processing horsepower when trying to actively monitor thousands of base-stations simultaneously. Existing routers and switches in the network aren’t (yet) architected for such a function. A number of carriers have proven, with catastrophic results, that their routers and switches aren’t able to handle such a load.
Furthermore and probably the most important point, Ethernet OAM only measures Ethernet. All of the services running over these phones are either IP based today or will be IP based in the future, even voice. Carriers I have spoken with have expressed a wish to monitor the IP and IP service layers as much, if not more, than the transport layer.
Performance Management
One-way measurements is at the top of the list of Carriers wishes for service monitoring over Ethernet based networks. While measurements on latency, litter and delay are all necessary Ethernet currently only measures round-trip performance. This limitation is one of the main reasons why mobile backhaul has taken so long to accept Ethernet as a transport technology.
In order to provide one-way measurements there needs to be some way of handling Ethernet in a synchronous-like fashion. Since Ethernet, by definition, is an asynchronous technology, this is no easy feat.
Standards bodies are reviewing all of the potential methods to make synchronous-like behaviors possible in Ethernet (1588v2, SyncEth) but we’re at least 1 year away from a real solution. That doesn’t address the fact that the millions of installed Ethernet ports in carriers networks worldwide lack the hardware to take advantage of any new technological advancements. I am not so sure that 1588 v2 or Synchronous Ethernet, in their current designs, will ever see mass adoption.
A number of carriers have proven, with catastrophic results, that their routers and switches aren’t able to handle such a load
Another equally important aspect of performance management is to use the measurements for trending purposes. Many issues in a network aren’t visible right away but rather creep-up over time. The ability to save measurements over-time and then apply policies and rulesets for interpretation can yield another perspective of the health of the network. This is especially important when the transport of the backhaul network is heterogeneous to highlight any long-term degradations that can be masked by digital services.
Traffic Condiditoning
When a service traverses multiple transport/physical technologies a number of issues can (and will) arise. These issues can manifest themselves as increased latency, jitter or re-transmission in the case of IP/data services. These issues cause degradation of service, inefficient bandwidth usage and frame-loss. One should also remember that there are differences between Fast Ethernet and Gigabit Ethernet that will give headaches if not allowed for.
These issues aren’t exclusive to media or protocol conversion but can also be traced to inadequately engineered hardware. Some base station vendors (who shall remain nameless) might be good at building radios but provide less than acceptable ethernet interfaces for backhaul. Then we must take into account that many mobile carriers around the world have many types and brands of base stations in operation.
Carriers are now asking for backhaul solutions that can condition the traffic between the wireless and wireline networks (i.e. backhaul) to ensure that none of the aforementioned artifacts (jitter, delay, disruption) are minimized, if not eliminated altogether. Conditioning of traffic cannot be limited to the transport or Ethernet layer. The IP and IP service layers must also be conditioned to transit the network as efficiently as possible.
In Conclusion
The MWC was a fantastic opportunity to talk with a number of carriers and vendors who all know that there is an issue but I have yet to see anyone who has a complete end-to-end solution. Some vendors have great radios, some vendors monitor excellently, and some vendors offer a cost efficient way of backhauling. As has happened many times before a very clever network engineer somewhere around the world will come up with a way to solve these problems elegantly without going way over budget and ensuring a consistently good revenue stream.
Are you that engineer 
?
