Networking Epochs from Mainframes to NaaS

The Important

  • After decades of Internet and Cloud infrastructure innovation / investment, the networking world turns its eyes to Enterprise Routing, and more generally, Enterprise networking.
  • The tip of the spear is SD-WAN, but Network-as-a-Service (Subscription) is the larger theme, with multicloud also part of the mix.
  • This presents Enterprise Networking disruption potential, though Cisco is not sitting by idle.
  • A new epoch in networking is under way. We look forward with great interest, and with more analysis to come.


The decline in the Service Provider Routing opportunity and the strength of interest in SD-WAN, has got us thinking about 1) why now and 2) the different networking epochs. After a couple of decades of Internet infrastructure investment/innovation, it seems almost natural the pendulum would swing strongly back to Enterprise networking. In this epoch, Cisco faces perhaps its strongest Enterprise challenge since the early to mid-90’s. That said, with two assets positioned in the SD-WAN bucket, it is far from asleep at the wheel. Still the question remains, will the Enterprise networking market be reshaped by an SD- WAN insertion, and ultimately by NaaS?

In the 70’s and a good part of the 1980’s, IBM dominated Enterprise networking with an array of Systems Network Architecture (SNA) and pre-SNA networking technologies. Very much shaped by the mainframe era, resources were allocated to, and managed by, a mainframe computer. In this epoch, the concept of Enterprise-style network management emerged, and transmission links evolved from low-speed async / bi-sync analog circuits to modern data link control based digital circuits. Routing was static.

This IBM dominance came under some pressure from mini-computers, but it was in the era of PCs, LANs, and later client / server that Cisco emerged from a pack of startups to dominate the multi-protocol routing Epoch that was quickly leveraged to dominate Ethernet Switching. The latter would become Cisco’s biggest business. Cisco leveraged general purpose processes and IP to drive this disruption. Routing was now dynamic.

By the mid-90’s the Internet was starting its meteoric ascendency and of course Cisco was an important player. But, the performance and scale requirements of the Internet were well beyond most, if not all , Enterprises. Into this void stepped multiple startups with Juniper Networks emerging in the late 90s and early 2000s as the cream of that crop. Juniper built its own ASIC-based network processor, popularized a modular, memory protected UNIX ( BSD ) as a base for a NOS, innovated in numerous other ways, including riding the MPLS wave.

The 2000s saw continued new entrants like Nokia (Timetra/ALU), Huawei (that was previously in the telephone switch business), and ZTE, carve out chunks of the SP routing market as IP/MPLS continued its drive from core to edge to aggregation and mobility. There were other import plots in this epoch and the previous one: BRAS, PPP, Frame Relay, ATM, and then Ethernet. Routing was dynamic, IP-based traffic engineering now an option, IP VPNs well deployed, the Internet robust, and broadband available.

With the Internet infrastructure scaled to support a significant amount of broadband, the focus of all IT shifted to data center / cloud. In stepped a new startup: Arista Networks. Low-latency switching, Linux-native routing, centralized / accessible state, and a value chain that evolved to be relevant to cloud / data center switching (with existing router vendors still having a good chunk of large cloud WAN routing).

So after a couple of decades focused on scaling the Internet and Cloud, what’s next? I am not saying that there was no innovation in enterprise networking between the 90’s and today, there was after all VoIP, Ethernet Switching, WLAN, secure routing, etc. but there was nothing that looked like it had the potential for a new entrant to rock Enterprise routing/networking with something that completely changed the game. Nothing until SD-WAN.

MPLS has been enormously successful, but it is not everywhere / on all access links, so it cannot always be an end to end virtualization layer. After decades of device-centric management approaches, networks are still manual and complex to manage with Cisco’s CLI and CCIEs having become defacto industry standards. As network conditions change we want to preserve and prioritize application-centric outcomes. Security remains a challenge. Why not leverage all this cloud and ML stuff to totally change the customer experience. And while we are reinventing networking, let’s take a look at subscription business models as well.

An epoch for mainframe computing, personal computing, internet-connected devices, and now, one for…the cloud / multicloud. Multicloud is part of this soup as well. Who will emerge with the most compelling vision of and execution for SD-WAN, NaaS, and multicloud?

After decades of one company dominating Enterprise networking, Enterprise networking suddenly seems fresh, full of new challenges and opportunities. Even Oracle acquired a SD-WAN company.

With Cisco having dominated Enterprise routing for so long, having done acquisitions in this area already, and with the financial capacity to do more, it is far from a slam dunk that this epoch will produce a disruptive new entrant, but the opportunity is there, and who knows, maybe the disruptor will be a (managed) service provider.

We will watch with interest as we analyze the SD-WAN / NaaS /multicloud epoch (more simply, the NaaS epoch).

This site uses Akismet to reduce spam. Learn how your comment data is processed.