SD-WAN and the dynamic nature of hybrid networks are making it easier than ever for CIOs to diversify WAN connectivity and rethink their IT infrastructure. But this new agility brings with it the responsibility to engineer the smartest network design. After all, IT architectures impact application reliability, visibility and security.
The problem with many WAN connectivity design approaches is threefold:
– Many designs fail to lead with business needs. All too often, the technology platform or access methodology dictates the design. Masergy performed a quick poll via social media and found that 78% of respondents agreed that this is a problem with network design.
– Many struggle when it comes to understanding when to use each type of access methodology and how to design around business needs, keeping price and performance top of mind.
– Many designs take a hardliner, homogeneous approach to connectivity, when instead hybrid networks that blend both private and public access methodologies might be the best option. You don't have to decide between an all-private or all-public network.
No doubt, there's both an art and a science to sound network design. The artistry comes with years of experience, but the "science" is really just a few design principles and best practices that are easy to learn. Below, we present a three-step process to overcome the challenges mentioned above.
Guiding principles: Price, performance and risk
Translating business goals into an ideal network design can be tricky for some. The key is to let business needs lead the design and then strike the right balance between price and performance, so the network adequately supports business continuity and doesn't cross the threshold of risk tolerance.
Avoid the common misstep of designing from a 'speeds and feeds' spreadsheet or from the limitations of a single, predefined access methodology. Instead, desired outcomes should inspire the design. Function should drive form – not the other way around. Start with an inventory of your apps, user groups, workloads and workflows, prioritizing the importance of each. Then design from there, keeping the cost and reliability of each access methodology in mind. Here is the process that I recommend, as well as some key considerations that can help keep designs founded in best practices.
Step 1: Prioritize your needs and understand connectivity pros and cons
Once you've prioritized your needs and are aware of the potential hurdles on the way, prepare with the following two pieces of information which should guide your strategy:
Start with your business continuity needs and chart your risk tolerance: In order to navigate compromises effectively, you should have an intimate understanding of your business continuity risk tolerance broken down by application, location, and user group and then also categorized by importance--as critical, important, or discretionary. The result of this inventory exercise should act as a framework for prioritization and a network design blueprint that allows you to match application/location/user groups with appropriate connectivity types.
How to do it: Create lists of each office or branch location, user group, and business application and categorize each as critical, important or discretionary.
Critical should be used to identify the items that are absolutely essential to your success. Many CIOs think about this in terms of revenue generation or production processes. Can your enterprise manufacture goods or make sales without this application/location/user group?
Important designates the items that are a priority but not necessarily critical. Would your enterprise achieve its goals without this application/location/user group?
Discretionary should be matched with the application/location/user groups that do not impact your business goals, such as guest Wi-Fi.
Review connectivity types and the relationship between price and reliability
Before attempting to match application/location/user groups with the appropriate connectivity types, review their availability risk and their price point. Ultimately, you get what you pay for. The more it costs, the more reliable it will be:
– Private access (SD-VPN): Higher price point but extremely reliable.
– Direct internet access (Public): Mid-tier price point and more reliable.
– Broadband internet access: Lowest price point and less predictable.
Step 2: Design your network
Now you're ready to design. Let your business needs and your risk tolerance be your guide in mapping your application/location/user groups to the ideal network access types that meet both your performance and budgetary requirements. Ask yourself: In an ideal world, what would each location, business unit and app use as a connectivity method?
Through the design process, you shouldn't feel pressured into abandoning existing private connections, but rather identify any discretionary locations and less-critical applications that may be able to transition from private to public (i.e., from MPLS to direct internet access or to broadband internet access). Remember to design freely, mixing and matching private and public connectivity types as needed.
Step 3: Draft plans and requirements for agility and visibility
Once you have mapped your ideal network, the next challenge is execution. Whether you're using a DIY approach or leaning on a managed network services provider, these technical requirements will make it easier to identify a networking plan that is flexible and agile enough to deliver on your ideal design.
Application and location performance visibility: Because not all of your applications or locations will require the same level of reliability and network support, your solution should provide you with deep visibility into performance as it relates to application and location. This will be fundamental in helping you optimize performance, reduce business continuity risk and deliver your desired design.
Network agility: Agility, flexibility, and ease of management are the characteristics of a single global network platform operating on software-defined networking principles. Before you sign a contract, make sure you are aware of the provider's infrastructure and change management processes.
Access agnostic: You should be able to mix and match connectivity types and last-mile vendors to strike the optimal balance between performance and price. An agnostic approach to access enables network designs to be customized based on your unique application environment, your user group and location priorities.