In a recent column, AvidThink founder and principal Roy Chua made the case that containerization and cloud-native functions (CNFs) have yet to truly arrive in telecom despite the hype. In particular, Chua likens the current cloud-native craze to the "NFV-washing" before it.
Indeed, a perusal of marketing collateral and press releases emanating from companies that offer any combination of cloud and containers often yields near-obligatory boasts of being "cloud-native."
Even some of the evangelists concede the point.
"Going from appliance to virtualized network functions was initially little more than putting that application into a VM (virtual machine) without much change to the design of that application," Ian Hood, chief technologist, global service provider at Red Hat, said to FierceTelecom. "Similarly, packaging that same monolithic application into a container is quite similar, as it does not achieve the gains of cloud-native application design."
For his part, Chip Childers, vice president of technology at the Linux Foundation's Cloud Foundry Foundation said that there are still potential cost benefits to cloud-native-branded "lift-and-shift" moves into containers, but that those benefits strictly take the form of "achieving the traditional goals of IT as a cost center." True, platonic cloud-native initiatives, said Childers, are yet more powerful.
"As with 'cloud-washing' a few years ago, we're going through another exciting round of vendors tacking 'cloud-native' onto anything that they can," said Childers. "When you make the shift to the right processes, architecture, and technologies, you unlock software as a competitive advantage."
According to proponents like Childers and Hood, the primary real-world advantage of true cloud-native approaches appears to rely upon speedier development and production cycles. Hood in particular points to "assorted research and surveys showing up to 66% faster application development cycles and 35% lower operational time per application, as well as up to 35% lower infrastructure costs." And, to be sure, these very figures are frequently cited by numerous vendors and service providers.
These numbers, however, appear to originate from a 2016 IDC whitepaper commissioned by Red Hat, describing customer respondents' results with Red Hat OpenShift. In an email, an IDC spokesperson noted that "the 2016 white paper would no longer be considered valid by IDC since it's nearly three years old."
By itself, though, this tidbit does not inherently demolish Childers and Hood's general point. A separate IDC survey from November 2018 found that while 50% of respondents reported that under 30% of their applications were cloud-native at the time of the survey, 75% of respondents predicted that more than 30% of their applications would be cloud-native within a year. IDC's analysts have gone further on the proverbial limb in the firm's 2019 FutureScape report (also published November 2018), in which they predict that 35% of all production applications will be cloud-native by 2022.
IDC bases this 35% figure, however, on "the combination of new-generation production apps and refactored legacy apps"—the latter vis-à-vis "phased modernization of legacy applications to cloud-native architectures" and "rapid maturing of vendor product portfolios."
This begins to sound a lot like the kind of CNF-washing that Chua described. For some of the same reasons that Childers cites, Chua advocates that "taking an open-source software-centric path" is better than leaving cloud-native development to standards bodies or vendor hype, noting strides made in this area by the Cloud-Native Computing Foundation (CNCF)—another Linux Foundation project, and sister organization to Cloud Foundry Foundation.
This past fall, the Linux Foundation announced that its LF Networking (LFN) was formally collaborating with the CNCF to foster the eventual migration of virtual network functions (VNFs) to CNFs.
Growth and interest in CNCF—let alone cloud-native functionality in general—is roaring upward, if a recent conference transparency report is any indication. In its December write-up of figures from KubeCon + CloudNativeCon North America 2018, CNCF reported nearly 8,000 attendees, with approximately 70% being first-timers at the conference. In a survey of the attendees, CNCF found that the top two reasons for attending were networking and the breakout sessions—seemingly demonstrating substantial interest in actual cloud-native topics and influencers.
Childers, for his part, reported that cloud-native open-source projects are seeing rapid growth in market demand and actual adoption—pointing in particular to an oft-cited figure that more than half of the Fortune 500 (along with most top telcos) currently deploy Cloud Foundry projects.
"We are approaching the peak of the 'hype' wave (of cloud-native)," said Childers. "What's important to understand is that if there has been sufficient adoption, the demand will continue to balloon well after the hype has died down."
Hood is more specific in his prognostication, saying that cloud-native demand will probably peak "a few years out"—in alignment with the concurrent scale-outs of 5G networks, distributed edge computing, and software-defined networks using AI or machine learning. Taken together, these three celestial bodies of next-wave technology will not merely usher in a foundation of CNFs, but rather demand it in advance.
"In a disruptive economy, first-to-market, or revenue, can be a key factor in capturing customers' mindshare and revenue. Thus waiting for the technology and standards to mature and develop will impact the net benefits," said Hood. "So, rather than wait for all of the technologies to fall into place, service providers should be getting started now on the perpetual evolution of their business to succeed in the disruptive digital economy."