A business is no longer just about connecting up all the resources within one building.
Companies beyond a certain size will almost certainly have multiple locations, which will require linking up as seamlessly as possible, so that they appear and function as one enterprise to employees in all premises.
But this process can be complicated by the heterogeneous technologies that will be required outside of the building, necessitating expertise that could be beyond the skills of everyday network administrative staff.
In this feature, we look at how Carrier Ethernet can greatly reduce this complexity, and facilitate the management of a network that goes beyond the bounds of the local area.
The wider perspective
A corporate LAN in a single building will use a relatively homogeneous infrastructure, both in terms of the cabling and the setup.
Most of the wiring will be category 5 or category 6 copper, with possibly a fibreoptic backbone if traffic requires it. This will support an Ethernet IP network in most cases, running through strategically placed switching gear to manage the communications between individual devices, and between these devices and shared resources such as email and storage servers.
There will also be switching gear serving departments and network segments. But the client devices, servers, switches, wireless access points and other devices on the network will all communicate over Ethernet, and any remote management tools will use this protocol too.
As a result of this uniformity, internally the network will present a single packet-based system. However, once you start thinking about connecting one building to another over any significant distance, all of this can go out the window, almost literally.
The Ethernet wiring could be fed into a specialised interface that will convert the packet structure to Frame Relay, ATM, or even legacy X.25. SONET/SDH could be another, more contemporary choice for large corporations. All of these have different frame structures to Ethernet, requiring the packets to be transformed at one end of the connection, and then transformed back to Ethernet when they reach their destination.
An Ethernet packet has a header and a payload, with the former transmitted before the latter, and then some CRC error correction code. But SONET/SDH interleaves the header and payload information, and whilst ATM, frame relay and X.25 also use packet-like frames, their structure is different to Ethernet.
An Ethernet Type II frame consists of a 14-byte header and a 4-byte CRC checksum, but the data payload can be anywhere from 46 to 1,500 bytes. An ATM cell, on the other hand, has a 5-byte header and 48-byte payload, whilst frame relay uses variable-size frames like Ethernet, but these will contain at least 1,600 bytes and often more, so they can transport multiple Ethernet packets. Frame relay also leaves the error correction to the packets it is transporting, rather than having this built into its own structure.
In other words, all of these systems will require the Ethernet data to be either aggregated or split up before it can be transmitted. The respective hardware interfaces will do this, but their optimal configuration will require specialist knowledge, both of the system used and how to optimize it for the kind of traffic mostly likely to be transferred from the LAN.
This is beyond a regular level of expertise in LAN configuration. A specialist will need to be employed to set the configuration up - and then re-employed when things go wrong or need changing - or extra staff with the necessary skills will be required. Since the protocol transformation entails a potential bottleneck on performance, optimization will be necessary here, too, in order to minimise the effect.
Expensive hardware, pricey expertise, and costly monthly fees all limit who can implement and manage these kinds of connections.
Ethernet all the way down
This is where using a Carrier Ethernet connection for wide area networking comes in. Although there are still many other technologies for hooking up your wide area network, including the ones mentioned above, Carrier Ethernet has been a rising star for a number of years now.
In fact, the bandwidth available to businesses globally from Carrier Ethernet already surpassed legacy systems in 2012, according to Vertical Systems Group. By 2017, it is predicted that more than three-quarters of global bandwidth will come from Ethernet, whereas in 2003 legacy services accounted for 93 per cent.
Although Carrier Ethernet will still require special interfaces to connect the LAN to the outside world, these are based around Ethernet technology, and therefore far less complicated. This makes them cheaper than their legacy alternatives, and their configuration is an extension of configuring regular LAN resources, although there are Carrier Ethernet-specific management methods here.
Adoption has been accelerated thanks to the Multiple Class of Service (Multi-CoS) capabilities of Carrier Ethernet 2.0, and the E-Access service type introduced alongside it.
The former allows a service to be provided that defines quality of service levels for different types of data. The latter means that a connection can traverse the cabling infrastructure of multiple vendors, whilst maintaining a single service level agreement. This has enabled a wider availability of Carrier Ethernet, since a company won't need to have the same provider covering all its locations.
But whilst these features extend the flexibility of Carrier Ethernet, it's the manageability that makes this type of connection particularly attractive as a LAN extension.
The Carrier Ethernet 2.0 standard includes a whole host of specifications for enabling and enhancing Operations, Administration and Management (OAM). These extend across an entire Carrier Ethernet connection, even when it spans the infrastructure of multiple vendors, and allow the administrator to monitor performance, diagnose issues, and then fix them.
For example, if there is a problem from a drop in service, the culprit cabling or setting can be pinpointed, and then a fix organised. The Service OAM definitions for Carrier Ethernet 2.0 are based on IEEE 802.1Q-2011 for connectivity fault management and ITU-T Y.1731-2011 for fault management and performance monitoring, making them part of wider international standards.
The Service OAM capabilities of Carrier Ethernet 2.0 are expressed in three main specifications from the Metro Ethernet Forum - MEF 17, 30 and 35. The MEF 17 standard sets out the overall framework for OAM, defining the high-level constructs used to model the necessary components.
It also covers the relationship between the Ethernet, Transport, and Application Service Layers as defined in the earlier MEF 4 standard, which is a generic standard for Carrier Ethernet. By providing the context for other MEF standards that define the Type 2 User Network Interface and External Network-Network Interface, which are integral to Carrier Ethernet infrastructure, MEF 17 also ensures that these pieces of hardware can be managed using standards-compliant tools.
This provides the ability to detect, verify, localize and receive notification of faults. It also enables performance monitoring, as well as auto-discovery of service-aware network elements within provider networks.
The MEF 30 standard focuses specifically on fault management, building on the MEF 17 framework. It defines how Maintenance Entity Groups can be set up to focus on a particular section or transport path in the OAM domain, based on the IEEE and ITU-T specifications mentioned earlier.
MEF 30 compliance means services conform to the necessary fault management standards. The MEF 35 standard, in contrast, is all about performance management, and defines specific measurement procedures as well as specifying solutions for collecting the information needed to compute performance metrics.
These in turn rely on the MEF 10.2 and 10.2.1 standards for Carrier Ethernet performance measuring, which allow precise service level specifications to be created. As we already explained, these OAM abilities extend both within a single provider's network and across multiple providers, so faults and performance can be detected even when they are on infrastructure that uses an E-Access connection. This ensures that the service level agreement can be guaranteed.
The net result of all these standards is the ability to provide all-encompassing tools such as Ciena's Ethernet Service Management platform, which puts all deployment and monitoring in one place.
However, there are also standards to streamline the implementation of Carrier Ethernet services in the first place, and these allow tools of this nature to enable users to efficiently build and deploy their Carrier Ethernet services. The standards for implementation of new services include MEF 13, 20, 23 and 26. The MEF 13 and 20 standards define how Type 1 and Type 2 User Network Interfaces are implemented, whilst MEF 26 (and its amendments) focus on the External Network Interface, for links between the infrastructures of different providers.
The MEF 23 standard, finally, is the standard for the Multi-CoS facilities, so the connection can be configured to prioritise specific types of data.
Simpler with Ethernet
There's no doubt that setting up, configuring, monitoring and repairing a WAN will always be more complicated than managing a regular LAN.
But where the world of legacy connections such as frame relay, ATM and SDH/SONET can be an alien one to everyday networking knowledge, a Carrier Ethernet deployment is at least based on a familiar technology for most systems administrators.
With a full set of standards for OAM, Carrier Ethernet deployment and management can be aggregated into centralised tools, greatly simplifying network administration.