The cloud has been one of the most popular buzzwords in computing for some time now, and it’s one that has plenty of reason to deserve the attention it has been getting. In fact, IDC has predicted that 2014 will see spending on the cloud surge by 25 percent, to reach over $100 billion, although this is still less than five percent of the $2.1 trillion overall worldwide IT spend predicted for the year.
However, the rapid growth of the cloud has had a knock-on effect on IT requirements in other areas, in particular the bandwidth and latency of a company premises’ connection to other company locations and the wider Internet. So, along with the growth in cloud, there is a growing need for faster connections. In this feature, we look at how an Ethernet WAN can play a significant role in satisfying that need.
Following the third way
The cloud is now often combined with social media, mobile computing and big data into the “third platform of computing,” and the Internet of Things is sometimes included here too. But its role in this quartet/quintet is not so much equal as fundamental, because the cloud is the technology that enables the other developments. The content of social media usually resides on cloud servers, mobile computing is vastly enhanced by ubiquitous access to data and files via the cloud, and big data is often hosted in the cloud as well. The key feature of the cloud is that its virtualization means the physical location of the data and services being accessed is subordinate to the ability to access them.
This is true of private as well as public cloud implementations, and perhaps even more so. A public cloud implementation such as storage or software as a service will be expected to be delivered from outside the local area network. But the private equivalents could be coming from servers in the building, over the corporate WAN, or even from some form of redundant balancing between the two. An employee will expect this to be irrelevant, and for the service to be responsive and usable, wherever in the company’s range of premises it is being accessed from. But in reality every service, even a redundantly load-balanced one, is hosted somewhere, and if that somewhere is accessed via a WAN connection rather than the LAN, its usability is going to rely on the potency of that WAN connection.
Traditionally, the WAN connections between corporate premises have used leased lines, frame relay or ATM. Leased line bandwidths range from 1.544Mbits/sec (T1) or 2.048Mbits/sec (E1) to 44.736Mbits/sec (T3) or 34.368Mbits/sec (E3), but prices can start at $1,000+ per month, and rise from there. So only the largest corporations have been able to afford the fastest versions, and even these are slow compared to the LANs they connect. With most desktop network interfaces now capable of Gigabit Ethernet speeds, a traditional WAN connection is likely to be orders of magnitude slower – tens, or even hundreds of times – which will be very obvious to end users. A Synchronous Optical Networking or Synchronous Digital Hierarchy (SONET/SDH) line is quite a bit faster at 155Mbits/sec, and ATM can reach gigabits per second by aggregating leased lines or SONET/SDH, but both are in the price range of telecommunications companies and very, very large organizations.
Allied with this is the complexity of these kinds of WAN connection compared to a LAN. With the latter, essentially the same protocol and technology is used across the entire network, from end users to backbone, even if the latter is fiber rather than copper. But where the LAN meets the WAN, a router switch will need to convert the Ethernet protocols to the alternative packet structures of frame relay or ATM, and then back again at the other end. The hardware to do this is expensive and requires specialized knowledge to set up. It also introduces latency and bandwidth constraints of its own.
Clouding the issue
All of this has made delivery of private cloud services over traditional WAN connections problematic. A June 2014 survey by the Cloud Industry Forum found that 78 percent of organizations had formally adopted some form of cloud service. The situation has been further complicated by the gradual shift from private to hybrid cloud, where internal and external services are woven together as seamlessly as possible into a combined whole. Some parts of a hybrid service will be coming from the internal LAN, while others will be arriving from the WAN or Internet connections. But only about 10 percent of businesses are looking at a pure cloud approach, according to the Cloud Industry Forum survey.
It’s clear, therefore, that a faster WAN connection at a much more affordable price than the traditional implementations is a necessity to take full advantage of the current cloud-oriented trends in IT provision. This is where the growing interest in Carrier Ethernet comes in. Where the traditional routes to high-bandwidth WAN connectivity are too pricey, and consumer-grade broadband technologies lack some important quality of service features that are necessary for corporate WANs, Carrier Ethernet has been specifically tailored for this kind of application.
While the latest version 2.0 specification of Carrier Ethernet has standardized the service even across different carriers, ever since its inception there have been some very good reasons why Carrier Ethernet has been the ideal technology for WANs. For a start, it is still essentially Ethernet, so a systems administrator who understands a LAN won’t need much extra training to cope with an Ethernet WAN. But right from the definition of Carrier Ethernet in 2005 by the Metro Ethernet Forum (MEF), it has clearly been specified with the scalability, reliability, quality of service and management attributes in mind that are suitable for a WAN, and Metro Ethernet services and the MEF existed for a few years before that as well.
The first iteration of Carrier Ethernet defined E-Line services for point-to-point connections and E-LAN services for multipoint-to-multipoint connections, with metropolitan and regional links. E-Line is ideal for connecting two branches, while E-LAN can connect multiple locations, with virtual versions of both allowing them to carry multiple connections across the same infrastructure, for example if the WAN serves a community of businesses sharing multiple premises in an industrial area. However, with Carrier Ethernet 1.0, the infrastructure needs to be supplied by a single provider, which limits its flexibility and reach. Premises outside the infrastructure of the provider won’t be covered by the service, and will need to rely on traditional WAN technologies instead.
Ethernet at the next level
Partly for this reason, in 2012 Carrier Ethernet was widened with some new specifications that formed the second generation. An important development here is the new E-Access service type, which allows a single Ethernet provision to run across the infrastructure of multiple vendors while maintaining its quality of service. This vastly expands the potential reach of an Ethernet WAN, making it a much more viable option.
The benefits can be considerable, too. Carrier Ethernet is already rolling out modules supporting 100Gbits/sec, with 10Gbits/sec and 40Gbits/sec already in wide use. Commodity services matching SONET/SDH, and considerably exceeding T3/E3 performance, are now available for pricing not far off what a single T1/E1 used to cost per month. While this might still be beyond the affordability of small businesses, and too expensive for use with small remote offices, medium enterprises and larger ones can now implement Carrier Ethernet at levels way beyond what was possible with legacy technologies. There are MEF specifications such as MEF 13, 20 and 23 that standardize the implementation of services, so providers can dynamically upgrade or downgrade a provision on request, making for a very flexible range of options.
Sheer bandwidth isn’t the only benefit, either. Around three quarters of companies are now using Multi-Protocol Label Switching (MPLS), according to Nemertes Research. This is because it can encapsulate packets from many different network protocols, allowing them to coexist. Ethernet fits extremely well into this. It can be carried over MPLS, so can sit alongside other types of network traffic that aren’t part of the WAN. However, the similar or even greater bandwidth available from pure Carrier Ethernet, and its reduced cost, are making it an attractive choice even here, with simpler deployment as well. The Multiple Class of Service ability added with Carrier Ethernet 2.0 means it can provide guarantees for latency-sensitive traffic that make it equally as viable as MPLS for many applications, and potentially more so when pricing is considered.
Ethernet power to the cloud
The cloud is still currently the concept that is exciting IT managers the most, and has been for a few years now. The cost savings and management simplification the cloud can have very real appeal. But the technology that can empower businesses looking to implement the cloud, particularly in private and hybrid forms, is very likely to be Carrier Ethernet. According to Vertical Systems Group, Ethernet bandwidth surpassed that of legacy services in 2012, and is set to contribute over 75 percent of bandwidth by 2017. This is why Infonetics has predicted that $150 billion would be spent on Carrier Ethernet over the next five years. So if you’re looking to expand your use of cloud technologies, an Ethernet WAN could provide the connectivity you need to make the strategy a success.
This article originally appeared on ITProPortal.com.