Carrier Ethernet's future may depend on smart apps, not speed

Carrier Ethernet services have been on the rise in recent years, but they're typically pitched as fast, cheap broadband connections to enterprises that can't afford ATM or leased lines. But vendors and operators alike are starting to move away from Ethernet's fast, cheap broadband pitch in favor of something with a little more value: optimized applications and services.

'When a service provider sells a data service to an enterprise, they ask the CIO if he wants 40 Mbps or 60 Mbps or something, and the CIO says 'I don't know, I just know that my application doesn't work',' says Andrew Coward, VP of service provider marketing and partnerships at Juniper Networks. 'So application performance is becoming more and more critical.'

That's especially true at a time when bandwidth revenues aren't growing in proportion to traffic, he adds. In short: 'Apps, not bandwidth, is where the money is.'

Service performance is also becoming crucial at the metro edge, according to Cisco Systems, whose new ASR 9000 edge router not only promises to boost metro network speeds into terabit territory, but also bundles service delivery functionality and video management into the mix. One heavily touted feature on the router is a content delivery blade that supports content caching, ad insertion, fast channel change and error correction.

'You can do your content caching at the local level, which not only improves video performance and help service providers meet their SLAs, but also helps save on international bandwidth costs,' says Sharat Sinha, director of_ service provider business for Asia Pacific at Cisco.
Cheaper content caching has its appeal, but Coward of Juniper argues that if you really want to put apps and services at the center of the Carrier Ethernet paradigm, you need to make the apps and the network aware of each other.

'Apps need service intelligence. We have to connect them to the business logic of how they behave on the network,' he says.

For MPLS champions like Juniper, that means flattening the network further so that IP-MPLS becomes the transport layer rather than the service layer. That in turn raises the ongoing argument over the future of Carrier Ethernet transport in the metro network based on either MPLS-TP or native Ethernet (via PBB-TE).

Anup Changaroth, Asia director of metro Ethernet networks for Nortel Networks, argues that MPLS-based Carrier Ethernet is still too expensive for metro networks, and that for now, price still matters.

'It's true that there's not a lot of intelligence there with PBB-TE. But the idea of adding intelligence to apps is a great vision, but for now it's all sizzle, no steak,' Changaroth says.
David Emberly telecom equipment research manager at IDC, notes that one trade-off of adding intelligence to apps running over Ethernet is that it changes the traditional Ethernet paradigm.
'Carrier Ethernet has always promised simplicity and low cost, but delivering intelligence simply makes things more complicated and makes it cost more,' he says.

On the other hand, he adds, that's not necessarily a bad thing. It could even prove a crucial differentiator for Carrier Ethernet vendors and their customers who can only compete so far on cheap bandwidth.

Related news and analysis:

 

About the author

Commentary

5G and data center-friendly network architectures

Matt Walker / MTN Consulting

Webscale and transmission network operators' interests are aligning as the 5G era dawns

Matt Walker / MTN Consulting

Webscale and transmission network operators' interests are aligning as the 5G era dawns

Rémy Pascal / Analysys Mason

The launch of 5G by South Korean operators serves as a first benchmark for other operators around the world