Sep 1, 2015
Most network organizations don’t have people in the organization who have a lot of free time on their hands. In addition, if there are some resources that can be freed up to analyze the possibility of introducing new functionality into the network, there are lots of competing candidates for what functionality to analyze. Given that, why do I advise my clients that now is the time to do a Proof of Concept (POC) relative to implementing a Software-Defined WAN (SD-WAN)?
While there are significant differences in the current set of SD-WAN solutions, most solutions separate the control function from the forwarding function and provide valuable new capabilities, such as the ability to centrally manage WAN devices, and to dynamically load balance traffic over multiple WAN links based on a combination of policy and the real-time characteristics of the WAN links.
There are many potential benefits that are associated with a SD-WAN. These include:
While each of the above are important benefits, I am going to focus on another potential benefit of implementing a SD-WAN: reducing the cost of WAN links.
In the traditional branch office WAN, connectivity into each branch is provided either by MPLS or by a combination of MPLS and direct Internet connectivity. Because of the previously mentioned capability to dynamically load balance traffic over multiple WAN links, one of the possible benefits of implementing a SD-WAN is the ability to make more use of relatively inexpensive Internet connectivity and less use of relatively expensive MPLS services.
I will use a simple example to drive home the potential cost savings. Consider a hypothetical company that has 100 branch offices, each of which uses 2 T-1 links to connect to an MPLS service, and each T-1 link costs $500/month. That branch office WAN costs $100,000 each month and $3,600,000 over a 3 year period. Further assume that the way that the company’s WAN traffic has been growing it is expected that within a year that they will have to add another T-1 access link to each branch office at a 3 year cost of $1,800,000.
If the company implements a SD-WAN, there are 3 alternative scenarios for how they could better manage their bandwidth costs. Those alternatives are:
In this scenario the company keeps all of its MPLS in place but doesn’t add any more to support the growth in traffic. The growth in traffic is supported by adding a high speed Internet link at each branch office at a cost of $50 per month per office. Adding the Internet connectivity has a 3 year cost of $180,000 but this cost is more than offset by the $1,800,000 the company doesn’t have to spend to add more MPLS.
In this scenario the company removes a T-1 access link from each branch office and adds an Internet link that runs at a higher speed than the Internet link that was part of the preceding scenario and which costs $100 per month per office. Adding this Internet link has a 3 year cost of $360,000 but this is more than offset by the $1,800,000 that the company saves by removing a T-1 access link from each branch office.
In this scenario the company eliminates all of the MPLS access links and implements high speed Internet connectivity at each office at a cost of $140 per office per month. Adding this Internet link has a 3 year cost of $504,000 but this is more than offset by the $3,600,000 that the company saves by eliminating all of its MPLS.
Ok, I freely admit that the preceding example is oversimplified. It does, however, drive home a point: the deployment of a SD-WAN has to potential to result in significant savings. In my experience, few things get the attention of senior management as much as when you can point to a line item, such as the monthly cost of MPLS, and tell them how you will reduce that line item.
Given all of that, why would you not do a proof-of-concept of a SD-WAN? In the best case, you prove to yourself and to your management that implementing a SD-WAN leads to significant benefits. In the worst case, the results of the POC are not compelling. However, just the fact you are doing a SD-WAN POC may be enough to cause your current MPLS service provider to offer to reduce the cost of their MPLS service in order to convince you to not implement a SD-WAN. Either way, you win.