Demystifying SD-WAN, Hybrid Networking and Their Impact on Business Performance
Hybrid networking and their impact on business performance
Software-defined WAN (SD-WAN) is the latest buzzword in the software-defined world we live in, offering all sorts of promises, but do we really understand the differences between SD-WAN & hybrid networking and what impact they may have on business performance and end user productivity compared to traditional private MPLS networks?
Read-on and hopefully I’ll help unravel this for you so you understand the key areas that may impact your organization. My aim is to help you make good choices, by being well informed.
- Hybrid network: a hybrid network can be considered a combination of different network types that can be combined to appear as one transport platform, irrespective of the underlying type (public or private) and transmission medium (eg. copper/fibre vs 4G-LTE/Satellite for example).
- In the world of SD-WAN when people refer generically to a hybrid network they typically mean using a combination of traditional MPLS (private WAN) and Internet services (public) to provide diverse connectivity to business applications, often with IPSEC tunnels being used over the Internet as an alternate WAN path.
- SD-WAN: the means of segregating configuration and control away from the data plane such that (in most circumstances) configuration is performed through a web browser in an intuitive GUI with little or no manual configuration being applied by humans to devices at the edge (the SD-WAN gateways), massively reducing complexity, lowering ongoing cost of management and improving business agility.
- Can I have a hybrid-WAN without a SD-WAN?
- Yes, absolutely, but you will likely have a mix of routers and firewalls at lots of sites, each with their own configuration interfaces or platforms—once you have more than a couple of sites connected in a hybrid WAN scenario the complexity of the configuration required to support a large mesh becomes very high and troubleshooting is a nightmare.
- Why would I want a SD-WAN when my traditional WAN (probably MPLS) does what I need?
- More organisations are adopting a wider range of applications and the delivery has moved away from on-premise to a combination of cloud-hosted, SaaS and on-premise applications.
- In a traditional MPLS WAN all traffic destined for the internet or public cloud is typically backhauled to HQ or a core data centre to exit centralized firewalls and then out to the application provider.
- If the applications are deemed business-trusted, such as Microsoft Office 365, SalesForce.com, SuccessFactors, etc., then the additional transit network impact of tromboning traffic over the WAN and then out to the internet, could add significant extra delays which result in poor end-user experience—in an SD-WAN hybrid network, trusted applications can break out, securely, to the Internet locally in each branch, improving the transit time to/from the cloud/SaaS applications, and reducing the load on the high quality, yet more expensive MPLS WAN. Freeing up capacity on the MPLS WAN provides greater bandwidth for other internally hosted applications.
- If I have a traditional MPLS WAN, then surely just adding more bandwidth would be a quicker solution to the performance problem?
- Think of bandwidth as a multi-lane highway, and you are trying to drive your car from Perth in Western Australia to Sydney, New South Wales.
- Adding bandwidth is like adding another lane, taking that two-lane highway up to three lanes. So unless the highway is already congested, adding more bandwidth won’t help—the two limiting factors are the distance (latency) and the top speed of my car (speed of light through fibre optic cable).
- My car has a top speed of say 150KmH, and it’s about 3900 K’s from Perth to Sydney—unless I can reduce the distance (latency) between the two cities, or increase the top speed of my car then adding the extra lane (bandwidth) doesn’t help. Add to this the top speed of my car (the speed of light through fibre optic cable) is a known maximum parameter, the only thing I can attempt to control to improve the trip time is distance.
- If I can reduce the distance (latency) between my user and the application they are trying to access by breaking out to the Internet locally in the branch then I have the opportunity to improve end user experience with that application.
So understanding all of the above, how can I best understand what is really going on with my network today to be sure that the changes I make will be the right changes for my business and give me the greatest bang-for-buck in business performance improvements?
Most organisations don’t have powerful end-user-experience (EUE) monitoring tools deployed across all of their network, so more commonly they will probably only have access to bandwidth utilization statistics on their WAN links. Add to this the fact that those bandwidth monitoring tools will also likely only show five-minute averages for the utilization, unless your links are flat-lining, you probably won’t have enough meaningful data with existing tools to gain the right insights to be well informed.
The best way to begin is by performing a true end-user experience assessment activity—pick say 50 or 100 of your users spread across different sites, that use a different mix of applications to ensure you get a good representative profile of your business. Use a comprehensive EUE solution, such as SteelCentral Aternity, to capture the required information over say 30 days, and from that data baseline your business.
Use the reports from this assessment to determine the real application performance and bandwidth usage profiles, per site, per user, per group, per application and get meaningful data on what today’s end user experience really is. With this data you can probably chunk your sites and users down into a number of typical groups, perhaps classifying your sites into type-1, type-2, ….type-5 categories and doing something similar for your business applications—getting a view on where the applications are served from, the typical high/medium/low end user response times today and with feedback from your users putting some science into what you really need to achieve to get tangible, worthwhile business productivity gains.
Knowing your business like this means you can develop a compelling business case that defines the goals, the investment required and the ROI it will deliver.
If you know this, you will know if SD-WAN alone will deliver the business benefits you need, or if in-fact you need a more powerful SD-WAN—at Riverbed we call this more powerful version AD-WAN—Application-Defined Wide Area Network.
Not all SD-WAN solutions are created equal. Riverbed SteelConnect is a true AD-WAN, a solution that is business focused, lets you define application specific policies AND determine which traffic will benefit from application acceleration and optimization using market-leading Riverbed SteelHead configured and supported from a web-based software-defined GUI. All of this can be delivered on a single platform—SteelHead-SD—the perfect combination of Riverbed SteelHead acceleration & optimization AND Riverbed SteelConnect SD-WAN.