What’s Old is New at KubeCon

What’s Old is New at KubeCon

Chris EvansCloud, Cloud Practice, Cloud-Native, Containers, DevOps, Enterprise, KubeCon, Opinion

I’m just back from a few days at KubeCon 2023, where I enjoyed catching up with old friends and making some new ones.  From the perspective of the CNCF project, it seems that what forms part of a mature operational IT strategy continues to be reinvented for the modern computing age.

The term “paradigm shift” is not one to be used lightly.  With respect to infrastructure, we’ve seen a few in Information Technology over the last six decades, notably from the mainframe to distributed/open systems, server virtualisation, containerisation and now serverless.

Each of these transitions is underpinned by advances in hardware that enables new features and functionality in software.  Monolithic mainframe architectures gave way to midrange computing platforms.  In turn, these were subsumed by the rise of x86 in the data centre. 

The slowing of Moore’s Law increased processor core count, requiring new thinking on how we multitask applications.  This enabled the container transition.  Arguably, the move to serverless will be driven by the ability to run processor-agnostic code on resource-efficient solutions such as Arm (and a range of new processors in the works).

Not all technology evolutions are directly data centre related.  The Internet, personal computing and mobile devices have all occurred in parallel to the paradigm shifts.  But those solutions are dependent on enterprise evolution.  If the mainframe was still the dominant enterprise technology, we’d all still be dialling in through dumb terminals today.


As I reflected on my briefings and impressions from KubeCon, I looked at the ecosystem and how new solutions such as Fermyon are addressing the same challenges we had with server virtualisation and containers.  In their case, it’s the addition of persistent structured data with a key/value store (more on this in a separate post).

Persistent storage is just one part of an ecosystem that encompasses aspects such as networking, platform, security, scheduling, observability and monitoring, structured data stores, code management and much more.

IBM Invented Everything

Going further back in time, all of the functions above existed in the mainframe days.  SMP/E was used for code management (generally the O/S, with other solutions for end-user code).  JES2 managed job scheduling, with time scheduling from OPC/A and CA-7.  VSAM (Virtual Storage Access Method) provided key-value, relative record and “time series” data sets.  DB2 and IMS delivered relational and hierarchical databases. DF/SMS provided persistent storage and data management.  Networking was implemented through SNA, while SMF and RMF provided observability and monitoring functions.  RACF implemented security policies. 

The aim of highlighting these tools is not to offer a history lesson but to show how each iteration in our IT evolution requires an ecosystem of solutions that deliver the operational maturity needed for their success.  There are fundamental aspects to all IT ecosystems without which we can’t operate effectively. 

VMware Reinvented Everything

Looking at server virtualisation as another example, VMware took the concept of the hypervisor (itself an IBM invention) and built the vSphere ecosystem around it.  This development included networking integration (extended with the Nicira acquisition), persistent storage (vSAN and shared storage support), monitoring and management tools (vRealize and now Aria), security (Carbon Black acquisition) and PaaS services.

The success of VMware as a platform (compared to other virtualisation tools) must be attributed to the development of the ecosystem, fuelled by smart acquisitions that addressed the operational aspects already known from the mainframe days.

Containers and Kubernetes

In terms of the containerised world, Docker started the journey that provided the base operating environment, while Kubernetes implemented a better scheduling and orchestration platform.  As we can see on the CNCF landscape, networking is addressed through CNI plugins, storage through CSI plugins with a range of solutions for security, code management, automation, and PaaS applications.

The CNCF Cloud Native Landscape is the ecosystem for containers.  Unlike the mainframe or VMware eras, most of the solutions are independently developed and not affiliated with one vendor.  However, the parallels with the past are still there, as CNCF looks to address the exact requirements we had in the mainframe and virtualisation eras but for the modern age.


What about the serverless world?  If serverless is to take off in a big way, then these same operational challenges need to be addressed.  Some pieces are intrinsically implemented by the nature of where the serverless code runs, whereas others will need “serverless-aware” implementations (like security, storage and networking). 

The Architect’s View®

What does all this mean for KubeCon, cloud-native and CNCF?  Unlike previous paradigm shifts, the cloud-native movement isn’t focused on one platform vendor.  Instead, we have many competing solutions offering similar capabilities.  If we parallel the current state of technology with geological time, we’re in the Cambrian Period of development with a wide diversity of solutions in our technology biome.  At the end of the Cambrian Period, we saw a mass extinction event, drastically reducing the diversity of species. 

In technology, this scenario is best described as rationalisation (or even standardisation), where a much smaller subset of technologies wins out.  In the 1980s, there was a diversity of home computers that eventually all converged on the IBM PC design (and clones).  In the data centre, x86 has become the dominant architecture, eliminating almost all the competition. 

We believe that the cloud-native landscape will also go through a rationalisation process as the category winners and losers emerge.  The market simply doesn’t need more than a handful of networking, storage and security solutions.  Code management has already been rationalised (as have some other areas).  We expect to see the winners emerge in areas such as data management, observability, service mesh, automation, and PaaS.

Like the standardisation already experienced in container runtimes, we should expect new industry or de-facto standards to emerge while software start-ups merge or are acquired.

In answering the question “What’s next for cloud-native?”, we see the standardisation and consolidation as a necessary step to the level of maturity needed for the cloud-native ecosystem to become the true next paradigm shift for IT. 

Copyright (c) 2007-2023 – Post #5cde – Brookend Ltd, first published on https://www.architecting.it/blog, do not reproduce without permission.