AI-assisted automation for clouds and networks climbs Gartner's hype cycles
Yes, AI will volunteer to run your IT estate. You might be a fool to trust it
Analyst firm Gartner has published its annual hype cycles for cloud tech and enterprise networking, and both suggest cooling your jets rather than rushing towards AI-assisted tools to operate your tech.
Hype Cycles rate expectations for emerging technologies, as they pass from being created as a the result of an "innovation trigger" and then ascend a "peak of inflated expectations" before often settling into a "trough of disillusionment" as they are proven less amazing than predicted. Technologies that make it out of the trough climb a "slope of enlightenment" before reaching a "plateau of productivity" upon which they are mature, widely applicable, and comfortably adopted by mainstream orgs. Technologies do not always reach the plateau – plenty become obsolete instead.
Gartner's 2024 Hype Cycle for Enterprise Networking rates AI networking and network AI assistants as having reached peak hype.
The former are interactive digital assistants that offer a conversational interface to offer NetAdmins what Gartner describes as "actionable network insights and assist with network operational tasks." Gartner rates them as immature, tied to specific vendors and therefore not useful on diverse networks, and likely to be ignored by networking pros who tend to be risk-averse – an instinct that kicks in when AIs offer suggestions without explanation.
AI networking also offers insights into how Net Admins can run networks, and Gartner feels the same reasons that make them leery of network AI assistants will make adoption slow. "Overzealous marketing creates confusion and makes it more difficult to select an offering that adds demonstrable value," the Hype Cycle adds.
AI Ethernet fabrics – networking hardware designed specifically for AI workloads – is a new addition to the Hype Cycle and earned its place on grounds that most orgs' initial AI adventures aren't on-prem, making task-specific hardware moot. A lack of standards is another problem, as are existing alternatives.
Gartner's 2024 Hype Cycle for Cloud Computing features "a flurry of activity at the Peak of Inflated Expectations." The analyst firm suggests that's to be expected given cloud is relatively new.
Generative AI-Augmented CloudOps – use of generative AI and machine learning to automate and improve management and support of public cloud services – is the most recent innovation Gartner has included. The firm rates it as potentially powerful, but warns admins could be "too quick to trust the output of GenAI tools." That's bad, because the firm's analysts warned: "Automation assets from GenAI systems are rarely production-ready. They can contain hallucinations and bugs in AI-generated responses."
The document rates AI supercomputers as a fine thing – albeit with uncertain ROI. Gartner therefore suggests using cloud-based AI supercomputing "for scoping resources to derisk and rightsize your investments."
Buyers are also advised not to rush into GPUs or other accelerators. "The latest CPUs with embedded AI acceleration can help train and infer models that are below 7B parameters," Gartner advises.
Assorted hype oddities
Among other items in the Hype Cycles that caught our eye is "coffee shop networking" – a debutant in the networking document. The term describes networks that just work across multiple locations, on an org's premises or in public. Gartner worries that it is "difficult to justify the shift to coffee shop approach if there are existing investments in next-generation firewall, SD-WAN, etc., that work fine and aren't due for refresh."
Another new entrant is Satellite to Device Direct (SCS) – a tech that combines satellite and cellular comms. Already present in smartphones that allow emergency calls to be made by satellite, SCS has obvious applications for remote sites. But commercial voice services for everyday calls aren't yet available and regulators are yet to approve the tech.
Hype for Wi-Fi 8 is also on the rise, but as the standards behind it aren't ratified and products won't arrive until around 2028 there's no rush to wrap your head around it. Gartner advises similar caution when considering Open Roaming – a network technique that eases provision of and access to guest Wi-Fi. The technique is little known, and creates privacy and privacy concerns you may wish to avoid.
- GenAI more buzz than biz as tech barely dents jobs, says survey
- Infosec teams must be allowed to fail, argues Gartner
- HPE picks up OpsRamp to sprinkle some AIOps on Greenlake multi-cloud
- IBM meshes with Flexera to boost AIOps IT automation
On the cloudy side, a tech closing in on the peak of inflated expectations is "headless SaaS" – software accessed mostly through APIs instead of a user interface. Gartner worries that "Best practices for pricing and procurement of headless SaaS are not well-developed" and that the absence of a UI means buyers will have to build their own – which could be a slow and costly process.
Gartner has declared cloud repatriation – the practice of moving apps on-prem from the cloud – obsolete.
"Even though repatriation hype is persistent, actual repatriation at scale remains nearly nonexistent and is not increasing," the firm found, adding "Repatriation hype often creates distractions during a time when thoughtful analysis should be devoted to appropriate cloud initiatives." ®