Packet Backhaul Busts the 'Pain Barrier'

LONDON -- Backhaul Strategies & Core Convergence for Mobile Operators -- Most mobile operators have overcome their fears that Ethernet backhaul would fail to provide good enough service quality, and many are rolling it out in volume now.

That's the message from Heavy Reading senior analyst Patrick Donegan, who kicked off Light Reading's Backhaul Strategies & Core Convergence conference here in London today.

This marks a major turnaround in operators' attitudes to and acceptance of packet backhaul compared with two years ago when this event was held in the same mod London hotel. [Ed note: which also happens to be the last known address of Jimi Hendrix, but I digress.]

At that time, the prevailing operator attitude to Ethernet backhaul was suspicion, mistrust, and lack of confidence in key areas. (See Carriers Don't Trust Ethernet Backhaul? and Carriers Go Slow on Packet Backhaul.)

But that is so 2008!

"Packet backhaul is through the pain barrier and operators are rolling out in volume worldwide," said Donegan.

After a lot of trials of many different architectures, solutions, and configurations, a number of leading operators are now operating packet backhaul efficiently and starting to get the better cost performance in the network in the volume that is needed to render mobile broadband services profitable.

So the waves of mobile data traffic crashing through operators' 3G networks, particularly in Europe, have swept mobile operators along to adopt packet backhaul as a cost-effective way to boost capacity in their transport networks to keep up with demand. Indeed, it is estimated that there is now three times more data than voice traffic on cellular networks in Europe as of this quarter.

It's this phenomenal growth in data traffic that is driving operators to deploy packet backhaul.

According to Heavy Reading research, 8 percent of all cell sites in Europe will have packet backhaul in live service at the end of this year, which is up from less than 2 percent at the end of 2008. Heavy Reading forecasts that by the end of next year, 17 percent of all European cell sites will have packet backhaul in live service.

In Europe, most of the packet backhaul deployments to date have been fiber-based, according to Donegan. But 2011 is expected to be a big year for packet microwave, and it will be deployed in substantial volume in "live commercial service," according to Donegan.

Out of sync?
The analyst noted that synchronization is still a bugbear for mobile operators when it comes to packet backhaul. A look under the hood of most packet backhaul deployments will reveal that operators keep an E-1 line in place for synchronization. (See Backhaul Timing: Anything But Synchronized.)

"There is still a lot of work to be done on operationalizing 1588v2 and [synchronous Ethernet]," said Donegan.

— Michelle Donegan, European Editor, Light Reading Mobile

tmmarvel 12/5/2012 | 4:16:15 PM
re: Packet Backhaul Busts the 'Pain Barrier'

The prolonged difficulties of synchronization with packet backhaul makes one wonder:

If the major costs of enabling broadband packet backhaul in any case arise from having to deploy fiber to the tower, then any cost differences arising from the choice of packet transport protocols running on the fiber should be comparatively quite marginal. In particular, the use of known working, established standard based--and natively synchronous--packet transport protocol, ie Packet-over-SDH/SONET (POS), should not break the budget. POS as Layer 1 can then be used to serve any L2/3 packet traffic, eg IP/MPLS. As such fiber with POS L1 to the tower should do it. 

The actual packet services in any case are L3 IP based, so the choice of L2/L1 protocols should be merely a tactical matter. Where synchronization is needed, the use of natively synchronous L1 would seem to make more sense.

torivar 12/5/2012 | 4:16:13 PM
re: Packet Backhaul Busts the 'Pain Barrier'

Because SONET/SDH is more complicated than Ethernet.  The hardware is just more expensive.   Most newer-generation gear supports SyncE so it becomes a moot point.  I work for a provider providing hundreds of cell backhaul T1s over packet networks built with SyncE with no issues.   There are no issues with SyncE as long as it's implemented properly and the devices support ESMC/SSM.  

1588 is more complicated and has a number of restrictions, but works over SyncE networks. Packet network congestion is the biggest issue facing 1588.  In our case the end to end path supports SyncE so it doesn't make sense to use 1588 even if it's available.  

scfm 12/5/2012 | 4:16:12 PM
re: Packet Backhaul Busts the 'Pain Barrier'

SONET/SDH is not inherently more complicated that Ethernet.  What makes it more expensive is that it has a much smaller end market and so all of the R&D costs have to amortized over a smaller volume.  Consequently, SONET/SDH (and telecom in general) need to operate at much higher margin levels to be healthy.

Carrier Ethernet, with features like SyncE and MEF/ITU defined OAM that are not required in 'LAN' Ethernet, will have the same market characteristics.


Sign In