alphaspirit - Fotolia

Get started Bring yourself up to speed with our introductory content.

How do you approach global MPLS procurement?

An expert discussed some of the particular challenges associated with the international procurement of MPLS VPN services.

There is no question that certain carriers have better in-region reach than their competition. That said, the actual provision and capability of a particular carrier's in-country presence requires some analysis to clearly understand the business impact. We therefore recommend creating a request for information (RFI), with questions about each region your organization desires connectivity. This process will force transparency from your prospective service providers.

The type of questions that you should ask include:

VPN provider edge (PE) -- The entry point into a global MPLS network infrastructure is known as the provider edge (PE). The PE represents the true capability of any prospective provider to service your locations. Telcos will often discuss reach without being specific regarding how your organization will connect to their actual PE coverage. The reason MPLS coverage is important relates to a number of critical areas.

MPLS network latency and jitter performance -- When reviewing the service level agreement (SLA) guarantee from network providers, the latency and jitter performance is calculated as an average over of a month (generally) and only refers to edge network nodes. In other words, the figures do not take into account your tail circuit. In order to understand the impact of tail circuit length on your applications' latency, understanding where PE nodes are located in your office site will allow estimates to be accurately calculated. When designing for global businesses, the success of application performance is clearly dependent on the overall latency. While a provider's network SLA might offer performance to support video and voice, a lengthy tail circuit may add too much additional latency -- resulting in an unusable application.

Of course, MPLS latency and jitter performance must also consider the prevalence and the advantages of cloud-based access. Normally, clouds are provisioned as regionally-based services -- for example, in the United Kingdom, United States, Europe and Asia. With cloud-based services, potential performance issues might arise if, for example, a European user is looking to access collaborative resources in Asia. Over and above performance, regulations within certain regions will also require special consideration, particularly China.

MPLS network resilience and platform diversity configuration -- There are a couple of ways to maximize connectivity uptime. The first surrounds the availability of dual PE devices for your network's primary and failover functions. An organization's primary circuit will route to the closest PE and then fail over to the nearest alternative PE. Keep in mind, the path to the secondary PE node will be longer than the primary, thus affecting latency. Once the availability of diverse PEs is determined, the next step is to confirm the capability of the facility within which the PE nodes are installed. We know of providers that operate highly resilient primary PE nodes but the secondary network of nodes is not as capable. Again, the RFI process will assist you to clarify these questions up front as you begin to implement strategy.

Multiple global network tail circuit providers -- National VPN service providers have the option to align their businesses with multiple wholesale local loop suppliers. This amalgamation, in turn, can add complexity to site migration and delivery. Let's use British telecom provider BT as an example. BT will only use one tail circuit supplier in the United Kingdom, Open Reach. When provisioning a global service, a single wholesale tail circuit provider isn't possible. And this is a major challenge. Wholesale local loop providers are all subject to different regulations and delivery workflows. That's why it's important for any prospective service provider to demonstrate its capability to manage regional challenges, timescales and any export or regulation issues.

Ultimately, a complete design that details exactly how dual local loop/last mile circuits will be configured is required to determine reliability.

This was last published in March 2015

Dig Deeper on WAN technologies and services