The notion of “network-as-a-service” or NaaS has been one of those things that’s regularly gained and lost visibility over time, while at the same time exhibiting a lack of a precise definition or even value proposition. In early 2023, I blogged about enterprise views on NaaS, and in the interval since then I’ve seen some evolution of the concept, including a July post this year in the TMF/Telco view of the topic. Where do enterprises stand now? I have some answers collected since the July post, that I think make some interesting points.
Since the July post, I’ve gotten comments from 88 enterprises on NaaS, fewer than I had for my 2023 blog because of the limited time. One thing that was consistent with the earlier view was the universal conviction of enterprises that the top benefit NaaS would have to deliver was lower cost. About a third went further, saying that cost savings was they only NaaS element they valued.
Another related and equally critical point is that 57 of the 88 users actually rejected the notion of universal usage pricing as a NaaS feature, and they all cited two reasons. First, they believed it would lead to higher costs, period. Second, they believed that NaaS support within their company was likely to come from line organizations, because cloud computing as a service was to them appealing to line organizations. They saw the democratization of network service procurement going that same way, and the way of “citizen developers”, leading some to characterize NaaS as a “citizen networks” technology.
Of the 88 users, 72 said that they believed that network service provider offerings of NaaS wouldn’t currently meet their requirements, but only 15 had actually gone through a NaaS proposal from such a provider. Of the 72, 44 seemed to have a generally cynical view of any new service offering, believing that what they were encouraged to buy was almost certainly what operators believed would earn them a higher profit. All 15 of those who’d actually assessed a NaaS offering had that more cynical view.
Since I’d written on this topic before, I did get some other follow-on comments. One was that it wasn’t just service cost that mattered, but “private network TCO”, including the cost associated with user-owned network gear and user-provided network operations support. However, enterprises were clear that they didn’t see themselves having network operators lease them LAN or other on-premises gear (84 of 88 rejected that), and only 45 of the 88 were at all interested in having the operator provide them with service-termination gear only. Of that group, though, 32 said they believed that operators wouldn’t be able to offer them the kind of deal on the CPE that they’d need.
One thing that I found interesting is that security, named as a NaaS benefit by 43 of 112 enterprises in 2023, was cited as a benefit by only 17 of the 88. More enterprises (22) thought NaaS posed a greater security risk. This shift seems due to enterprise efforts to harmonize their disorderly, multi-layered, complex, security environment. Adding a new player, it seems, is against the management mandate for security containment. In addition, they believed that the populist model of NaaS usage, which IT/network professionals smacks of “citizen networks”, would lead to amateurs assessing security, and thus almost certainly increase risk.
In 2023, less than a third of enterprises had any notion of how or whether SD-WAN might improve security, to the point that they could even list an SD-WAN vendor with security features. Of the 88, interestingly, nearly half understood how SD-WAN or any form of virtual networking could aid security, and these were largely the enterprises who rejected the notion of a NaaS security benefit. This group believed in the concept of virtual networking overall, too, and that was a topic that in 2023 generated almost no user interest or even awareness.
My view is that service-driven and even product-driven thinking and planning have had a decisive impact on the concept of NaaS, not to mention excessive operator and equipment vendor opportunism. Another factor, one that reinforces the latter of these impact factors, is the rethinking of the cloud we’re now seeing. The cloud essentially launched the as-a-service concept, and the cloud was (like, let’s face it, most technologies these days) overhyped and oversold to the point where backlash was inevitable. Companies don’t always cite it (only 27 of the 88 mentioned it this year) but I think a lot of the falling from grace we see NaaS experiencing is due to the fact that “aaS” itself has fallen from grace.
On the other hand, enterprise comments on NaaS and on how it might be subducted into a new vision for virtual networks makes me wonder. Are we struggling not only to name a virtualization paradigm in networking, but to define it? Right now, it appears that network virtualization is trapped in lockstep with compute virtualization and cloud computing. My July blog points out that cloud and multi-cloud is the focus of at least some of the new announcements in NaaS, which suggests that the linkage between virtualization in networks and that of computing is tightening. I think that’s a bad thing, not the least because the enterprises who I talk with are more skeptical about the value proposition of cloud and multi-cloud. I think that the premises side of virtual networks needs to be addressed, and neither the cloud providers nor the telcos can do that.
What should happen now is that virtual networking as a concept, a concept that then makes things like SD-WAN and SASE merely partial implementations, may now be the way forward. Virtualization created the cloud, enabling efficiency, scalability, and agility that something hard-rooted in physical infrastructure could not deliver. VPNs brought some of that to networks even before the cloud, and virtual networking forms an important (though usually unrecognized) piece of the virtual data center we’ve already committed to. Could it, should it, be our next transformation. I think “Yes!” Whether it will, though, is likely to depend on network vendors and how they present it.