Happy New Year. I hope 2005 finds you refreshed and ready for what promises to be an exciting year for MPLS. I hope everyone gets a chance to work with this exciting technology, I know I enjoy it immensely. Anyway, as in most of the articles I write, the information focuses on areas that an enterprise customer needs to consider when deploying an MPLS solution. This article will focus on redundancy between a customer site and the MPLS cloud.
I have worked with production MPLS for over 2 years now with multiple carriers and enterprise customers and the technology is sound (although the implementation of the technology may not be). We know that MPLS can provide solutions for convergent applications across a single IP interface. Now, more and more enterprise customers are looking at how to provide redundancy and robust connectivity from the CE as they move forward with mission critical traffic such as voice and video over their IP backbones. The days of ISDN and dial back-up are gone as customers are looking at broadband capabilities for survivability at their remote sites. This requires either dual access into a single CE or dual access into dual CE routers. I want to focus on the hidden concerns when making this type of decision regarding CE-PE resiliency options.
First of all, determine how critical the traffic is that will be traversing the MPLS WAN backbone. For any solution, the estimated maximum downtime can be determined from SLAs offered by the MPLS providers and WAN CE vendors. Either way you should determine what the mean time to repair (MTTR) is for the WAN link and gauge whether or not your application traffic can go that long without service. VoIP traffic can be diverted to the PSTN, but the data traffic cannot. If it is decided that there must be a secondary link, the link needs to be right-sized. This is where the secondary options for the redundant link come into play. The following questions will need to be answered:
- What access options are available from the provider?
- Do all applications need to be survivable?
- Can a smaller sized pipe be utilized?
- How does the traffic fail-over?
- Can the WAN CE support different access types?
- What routing changes do I need internally to ensure proper failover on the WAN?
- How do I interface with the MPLS backbone from a routing perspective?
These questions can be answered rather easily, but assessing the cost, impact and necessary changes required to implement the desired solution can be daunting. The routing from the customer's internal network to the WAN edge could change significantly if dual homing is deployed. This can cause the internal network's upstream to require redesign and reconfiguration. This is especially true in organizations that have core, regional, national, and international routing domains.
The key message here is that dual homing introduces more than just a decision process regarding access feeds and speeds. Dual homing can require routing architecture changes that can extend into the internal network. The providers generally do not provide internal network design, although most will provide input. Experience has shown that the decisions and tasks usually fall on someone on the client's staff. If that person is you, please try to consider all of the above when evaluating redundancy options with MPLS services.
Robbie Harrell (CCIE#3873) is the National Practice Lead for Advanced Infrastructure Solutions for SBC Communications. He has over 10 years of experience providing strategic, business, and technical consulting services to clients. Robbie resides in Atlanta, and is a graduate of Clemson University. His background includes positions as a Principal Architect at International Network Services, Lucent, Frontway and Callisma.