Cloud computing has transformed how modern businesses operate, enabling unprecedented agility, scalability, and innovation. However, effectively leveraging the cloud requires careful planning and strategic design principles. In this fascinating episode of the Altitude Podcast, host Woody chats with Rob DeWeese, Director of Cloud Networking at Kyndryl, to unpack cloud networking design principles that pave the way to success.
Here are the highlights from the episode.
Breaking Down Cloud Networking into 4 Towers
Rob shares his simplified understanding of cloud networking by breaking cloud networking down into 4 towers:
- Tower 1 – Hyperscaler Land: Anything and everything within the hyperscaler in every possible direction – east, west, north, or south.
- Tower 2 – Hyperscaler WAN: How we connect in, be it through the internet, VPN, MPLS, etc.
- Tower 3 – Interhyperscaler and Service Control: The arsenal of tools responsible for regulating and directing traffic, which can include routing, security, management platforms, and cost control.
- Tower 4 – Complex Flows: SaaS applications, APIs, and future-proofing for expansion, integration, compatibility, and elasticity that extends beyond the cloud and is unrestricted by geography or circuit type.
Working from an understanding of these 4 towers gives a framework for design that encompasses the needs and demands of all the key stakeholders that rely on your cloud network.
Design a Future-Proof Network Through a Modular Approach
As a starting point, Rob advocates for designing networks in a modular way: “Design the network as a product built to adopt new features and technologies… Build it like LEGO, so you can bring in new technology and chuck things out.”
This approach future-proofs the network as you are able to adopt innovative technologies and drop what is no longer needed – without disrupting your network.
Build Your Network to be Transport Agnostic and Independent of Connectivity
Once you’ve achieved agility through modular design, Rob also suggests building a design that works regardless of circuit type or geography, highlighting the importance of bringing your service as close to the end user as possible.
He shares that the network should be designed to be “transport agnostic and independent of connectivity used because nowadays, it’s about enabling your platform to work however people want to consume it.”
Avoid Vendor Lock-In for Core Networking Services
As a final step towards agility, Rob shares the importance of avoiding vendor lock-in for core networking services and uses Aviatrix as an example of a vendor-agnostic solution: “Avoid cloud vendor lock in for core services that run the environment, which is where Aviatrix fits in.”
He shares that businesses in multiple clouds don’t want to have an Azure firewall expert or an AWS firewall expert and run Palo Altos on prem. Businesses are thinking about speed to execution and management, so “ease of administration and complexity really matter.”
Businesses need a cloud networking platform that provides a single solution to manage and secure infrastructure across AWS, Google, Azure, Oracle, and Alibaba.
Interested in hearing the rest of Rob’s tips? Listen to the full discussion.
Be sure to subscribe so that you don’t miss out on more insights from real-world practitioners, industry analysts, and thought leaders from leading companies such as AWS, Microsoft, Equinix, and Kyndryl. It’s not just about staying informed — it’s about staying ahead.
Get your hands on A Network Engineer’s Guide to Multicloud Networking today. Multicloud networking shouldn’t be complex. Simplify it.
