case studies

2011 COMPTEL PLUS Las Vegas

join us!

request for info

Connet China
interviewed by HKET

Frame relay is a connection-oriented protocol and well suited to a simple data switching infrastructure, unlike connectionless protocols such as IP.

Network comparison of 6 sites

For some users, you might actually be able to reduce hub site bandwidth requirements vs. those of a frame relay network, because of the more dynamic nature of any-to-any connections. But at the same time, bandwidth requirements may need to be broken down by CoS, and CoS throughput bandwidth definitely will need to be included at hub and remote sites, taking into consideration application traffic utilizations and traffic flows. That’s particularly true for enterprises that currently employ partial or full frame relay PVC meshing, where the enterprise has multiple server farms/datacenters or requires traffic to reach anywhere because of any-to-any applications such as VOIP. Thus, the carriers may need to know more about your applications and performance expectations than they have in past procurements of new networks. And the more you can define those according to the IP/MPLS specs they can recognize, the better.

So you need to investigate with each provider the MPLS network’s physical and logical topology; any Layer 2 access used; whether the MPLS traffic is transported over the providers’ shared router public Internet IP or non-shared IP infrastructure; and their MPLS network bursting policies, which may be far different (and potentially more expansive) than their traditional frame relay policies.
Additional questions that typically arise include any enhanced MPLS services that are available, such as network-based firewalls or Internet datacenters, and the routing protocols supported on your customer premise equipment. Hint: Be prepared to implement eBGP on your customer premise routers!

All of this will affect your ultimate service level agreement with your chosen MPLS carrier. With MPLS, it’s easier than ever for carriers to slip into the habit of only tracking their performance within the carrier cloud, since MPLS label tagging is performed at the provider edge (PE) router rather than the customer edge (CE) router. SLAs should include site-to-site, and not just PE-to-PE router metrics, and should be based on site availability, not “network” availability. Here’s where styling your procurement project as not strictly MPLS can help. In some regions of the world, MPLS still is not either the most economical nor most popular choice. In Canada, long-distance, point-to-point Ethernet services have really taken off. And in China, historically, nothing has been cheaper than private lines. The reality, however, is that enterprises want to create pan-Pacific regional MPLS networks, and many carriers are creating MPLS NNIs with one of the two principal Chinese carriers. Letting the carriers know you are ultimately focused on the end-goal of an efficient IP network, not a specific carrier solution, will help sharpen up their responses to you.

The move to MPLS can help in this process. In the complex world of IP voice and data, there’s a premium to being able to present to current and potential service providers your ideal network and what it has
to accomplish in terms of applications, and at what kind of price. That means getting your own senior managers on board a plan, so that the service providers don’t wind up presenting you with options that move you to a next-generation network without actually saving you any money, or even wasting bandwidth. The reality of a looming change in data-networking generations has a way of concentrating minds. Each enterprise manager has the power to decide whether it will be a headache or a boon to their own competitive position vis-à-vis the few remaining competitive carriers across the nation and around the world.