Contributed By

john-hawkins

John Hawkins

Senior Advisor, Product and Technical Marketing of Ciena

View Profile

Why Ethernet for Infrastructure?

May 06, 2015

Ethernet as a data networking technology has been in wide use for many years, keeping pace with the network speeds demanded by applications and adding key service management and quality of service (QoS) attributes needed to succeed beyond the local area network (LAN).

Part three in a three-part “Improving the Enterprise Network” blog series addressing how private Carrier Ethernet can benefit the campus network, and the specialized WAN. To read part two, click here.

Ethernet as a data networking technology has been in wide use for many years, keeping pace with the network speeds demanded by applications and adding key service management and quality of service (QoS) attributes needed to succeed beyond the local area network (LAN). But considering it for a starring role in wide area network (WAN) infrastructure requires regarding the network from a different perspective.

Simplicity

The nature of the access and aggregation infrastructure lends itself to more stable, longer-term connections that are intended to backhaul user traffic to an edge node where information is held and processed (typically a data center— private, public, or some hybrid). This so-called “user-to-content” connectivity benefits little from full-scale routing and the complex and often proprietary functions that can overburden today’s router hardware. All that is needed are simple connections that can be traffic engineered and protected, if necessary, and provide deterministic (well-bounded) behaviors such as latency, jitter, and packet loss. The real purpose of the access/aggregation network is to connect users to their desired content or application, while making the most efficient use of the expensive fiber plant.

Network Performance

Because it operates under a uniform protocol from LAN to WAN, Ethernet connectivity avoids unnecessary protocol conversions and is ideally suited for a wide range of critical enterprise applications such as latency-sensitive storage applications, financial trading programs, critical infrastructure protection, and rich media applications. Combined with modern packet-based traffic engineering technologies such as MPLS-TP, specific routes (and backups) can be engineered for a given level of performance. These can be dynamically defined, but, often in the access/aggregation scenario, statically created tunnels will be preferred where deterministic performance is required. Also referred to as “connection-oriented”, these tunnels can be planned ahead of time, monitored on an ongoing basis, and adjusted to reflect dynamic bandwidth demand.

Network Security and Control

The connection-oriented Ethernet approach also ensures traffic is only delivered where it should be. Spanning Tree Protocol is no longer used, as the deterministic tunnels do not require the traditional learning or restoration functions performed in a traditional Ethernet LAN. Snooping of traffic is therefore less of a concern, as an inherent layer of security is built in, with layers of visibility and control of the Ethernet Virtual Circuit (EVC). Scalability is provided architecturally by using virtual switches within physical switches to provide secure, end-to-end traffic separation. End-users can control their own network assets without interfering with other users or the larger enterprise infrastructure.

Flexible, Scalable Bandwidth

The increasingly cloud-centric networking environment can benefit from sharing costs among a community of users. New software-defined networking (SDN) approaches have enabled scheduling of bandwidth on an as-needed basis for cost-sharing purposes. Dynamic, granular, bandwidth-on-demand avoids overbuilding the network and brings such cost-saving options to the table. Dynamic scalability—from 1 megabits per second (Mb/s) to 10 gigabits per second (Gb/s) and more— allows for adjustments (up or down) to capacity deployment on a per-site basis, either via a user portal or automated machine-to-machine interactions (such as a virtual machine requesting more bandwidth from the network). While these capabilities are not unique to Ethernet, the determinism and operations and maintenance (OAM) tools available to L2-based infrastructure makes it a compelling approach to providing such elastic bandwidth for end-users and applications.

Ethernet OAM

A number of standards-based OAM tools provide advanced means to monitor and manage the communication on Ethernet virtual private networks (VPNs). Again, while these tend to be carrier-driven features, end-users will also appreciate the level of visibility and control they provide in the enterprise environment. Tools like loopback and continuity checks help ensure a connection is up and running before it is relied on in a working environment. Others, like periodic latency and jitter measurements, help organizations ensure they are receiving the service levels they need for the health of their business applications.

The Bottom Line

Today’s network traffic flows and dependencies have changed due to the increased use of cloud and exploding bandwidth requirements. Enterprise network managers need to break free of business-as-usual approaches and take advantage of new tools for the network. A single, familiar Ethernet interface enables convergence of all services over a common network infrastructure and a unified end-to-end protocol, simplifying operations and taming the need to respond to escalating demands with additional network complexity.

This article is available exclusively to
Comcast Business Community Members.

Join the Comcast Business Community to read this article
and get access to all the resources and features on the site.

It's free to sign up

OR

Join the Discussion

300 Characters Left
comcast-ctas-ethernet-banner-315x289-v1