When spiders spin a web, they have just one goal – to catch insects. And with just one goal, they can keep things simple: one material, one construction method. But large companies don’t have such an easy task when it comes to building their networks. Multi-location business networks (WAN) don’t serve just one purpose and don’t connect just one type of user or infrastructure. Data centers need high bandwidths, smaller branch offices need low bandwidths; IP telephony requires short transmission times, but for email that’s not so critical. In addition, digitization produces a steadily growing number of applications and thus greater demands on the company network. Unlike with the spider web, one material alone is not sufficient to meet these requirements.
Two examples: Until now, an automobile manufacturer has used primarily optical layer 1 or layer 2 connections to link their data centers with large subsidiaries via broadband. But to network connected cars, they now require multiple encrypted Internet connections (layer 3) in their network. Or a large retailer: They have used cost-efficient Internet VPNs until now, but need a high-performance line to add cloud services to their business network.
Multilayer networks – networks and characteristics
Wanted: Improved quality and profitability
Added to this: The volume of data traffic in company networks overall is growing. But operations still need to be profitable. No question – digital business models need network design 2.0. The networks of the future don't need just one or two technologies, but a comprehensive combination of technologies that provides optimal connectivity for every application – in other words: a multilayer network.
Hybrid networks on the third level of the OSI model, the network layer, are an initial remedy: For access to the company network, cost-efficient Internet VPNs supplement the proven and powerful Multi Protocol Label Switching (MPLS) connections. If bandwidths or specific transmission properties of the layer 3 services are not sufficient, Ethernet services (layer 2) as copper or fiber optic connections are the next step. The growing need for high bandwidths has driven the use of Ethernet in WANs in the past years. The reason: With bandwidths of 50 megabits per second (Mbps) or more, Ethernet is often more cost-efficient than MPLS (see also White Paper: Hybrid Networks).
Ethernet over all-IP infrastructure – a substitute for MPLS?
Another argument in favor of Ethernet will be relevant in the future. This is because the all-IP infrastructure will soon enable specific transmission qualities (class of service, CoS) for many Ethernet services. Telekom will provide the majority of Ethernet services in this manner in the future. In addition, it will be possible to provide multiple logical networks (Ethernet virtual private lines, EVPLs) through one single Ethernet connection. Specific CoS profiles for latency, packet loss and jitter can thereby be defined for each logical network – a feature that was reserved exclusively for MPLS networks until now.
Easier network configuration
What does this mean in practice? Until now, for every separate VPN, a company had to have either a dedicated connection or configure complex separate VPNs to prevent their data transfers from interfering with each other. An example: So far, a retailer with multiple stores had one line each for the phone, the ERP system, the EC cash terminal and many more other specific applications. However, with the new all-IP based Ethernet service from Telekom, only one single network connection is needed for more than 4,000 logical connections and all the connected applications – as long as there is sufficient broadband. This makes Ethernet even more attractive than WAN as an access variant – as an alternative or supplement to WAN connection via MPLS as well.
Multilayer networks – networks and their use cases
But it also has limitations: Unlike with layer 3 services, companies that use Ethernet for office communication, for instance, need to take care of their own IP addressing. In addition, Ethernet over all-IP doesn’t reach the high bandwidth and low latency of standard Ethernet services in all instances. However, the latter methods will also be available on a new, optical platform in the future. Companies that use multilayer networks will thus choose between two Ethernet platforms depending on their requirements:
Ethernet over All-IP: short latency, high bandwidths, strong security, specific quality through CoS profiles
Ethernet over Optical Transport Network (OTN): even shorter latency, higher bandwidths (up to 100 gigabit per second)
The best network for every application
The segregation of duties of a multilayer company network could look like this: Data centers communicate with one another and with the corporate head office through managed layer 1 services via fiber optics. Companies can link medium-sized locations or specific cloud services using Ethernet over the all-IP platform (with QoS). A majority of site networking and therefore applications such as IP telephony will continue to be offered via MPLS (layer 3 with QoS), while Internet VPN will connect mobile devices, the mass of IoT devices and smaller locations to the company network. With this colorful network, companies are prepared for every digital scenario – and don’t need to shy away from comparisons to a spider web when it comes to load capacity.
If companies switch to SIP trunks, they should consider a PBX (private branch exchange) and a centralization of all connections as well. It is a decision that will save them more than just voice channels.