Cloud Networking Needs More Than Incremental Innovation

If you’re looking for a better way to build your network, everyone knows you go to ONUG.  ONUG has proven that community collaboration transcends convention and myopic perspective.  ONUG has been re-inventing what’s possible with the OSI stack allowing us to continue our forward progress when we were about to run out of road.  Ten years ago, I remember an event where the presenter terrified everyone (who mostly showed up for a free steak) with the “fact” that we were about to exhaust the public IPv4 address space.  I can’t say that ONUG played a part but, outside of ISPs, I still don’t see a lot of IPv6 even in 2022.

Out of the many topics covered at ONUG’s April session, the problem the Enterprise Cloud Community is looking to solve is essentially the same problem I took the “steak bait” to learn about…how do we scale IPv4?  Scale was the birth of cloud computing: “I can’t my get servers racked and stacked fast enough, I’ll use someone else’s.”  Now we are finding the same to be true for “Multi-Cloud Networking” (MCN).  Why build out a POP or COLO when you can use someone else’s?  Racking and stacking becomes their problem but scaling IPv4 still remains your problem.

Before everything started circling the drain in early 2020, it would be enough to show up to ONUG and talk about how you could automate routing or NAT policies as a way of making your expansion into public cloud less painful.  It might be enough to orchestrate virtual appliance deployment and configuration.  It sounds good, but is it just an incremental fix that only serves to kick the operational can down the road?  You don’t have to worry about IPAM, but you still kind of have to.  Deploying your virtual appliances was automated, but there’s still an operational bill due when you start to exceed capacity.  The automated orchestration sounded great until a critical CVE forces you to patch hundreds of virtual appliances that will likely – no surprise here – require a full reboot.  You don’t want incremental innovation in the cloud if it leaves you living in the same, operational house of cards.

Bespoke CNIs are pushing the efficiency of public cloud past what is possible in hypervisors, yet few organizations have fully committed to service mesh architectures within private cloud.  While this kind of innovation may be limited in production environments, it shows us the future of networking.  Present-day problems, however, remain.  While your strategy may include moving towards a service mesh architecture, it’s likely that it is only one part of a larger mix of cloud-native services, virtual machines, and those “just won’t die” legacy applications in the dark corners of your datacenters.  If you have made the commitment to a hybrid cloud strategy, incremental improvements to conventional networking are not going to deliver the goods at scale.   Even a legacy application running on virtual machines can be set to auto-scale, but monolithic architecture keeps network appliances from scaling up and down in the same manner.  

The CSPs have provided us with multiple options to facilitate connectivity to and within their infrastructure.  This is awesome at first but, at scale, visibility and consistency are easily lost.  Single cloud complexity is easily compounded by adding a second or third CSP even if we operate those clouds as islands as opposed to a cohesive fabric.  Regardless of their intent or your desired outcome, the needs of the business are rarely served by siloed or single-vendor implementations.  We need the ability to choose but it shouldn’t be a painful choice, nor should we have to resort to legacy constructs that degrade the value we get from leveraging multiple clouds.  We need the same flexibility, scale, security, and experience that we get from individual CSPs, but implemented consistently across all of them.

The term “next-gen” gets thrown around a lot.  Is a next-gen firewall still considered “next-gen?”  I believe that if anything’s truly next-gen, it’s the next-gen hybrid cloud network.  Going beyond just connectivity to the cloud, but to connect all clouds and their things – at scale, securely, and with performance – is what makes Multi-Cloud Networking truly next-gen.  Easy to say, harder to do.  If you and your organization choose to pursue MCN, incremental innovation may get you through Day 0, but the ROI for Day N+1 may have diminishing returns for Ops while adding risk to business objectives.  If MCN is your future, will incremental innovation be enough to get you there?

If this blog resonated with you, be sure to check in with us at the ONUG 2022 Spring Event. Prosimo will be conducting an Innovator POC on April 27th then delivering a Luncheon Keynote Address on April 28th. We look forward to seeing you there.

Author's Bio

Benjamin Molnar

Prosimo, Inc.