As I interpret the news, there are two reasons to move from cloud to premises:
1.- Security, the cloud is more vulnerable to attack so you take strategic, security sensitive apps back to your premises.
2.- Latency, apps that require massive data transfers suffer the most from cloud latecy so you take AI apps back to your premises.
In my view, neither reason is completely clear-cut.
Yes, you can make your private data center more secure than a public cloud, but not all companies provide the financial backing and employ people with the knowledge to do so. As we’ve seen, the private infrastructures of big companies with purportedly professional IT departments like Target and TJ Max were successfully hacked (as have dozens of other private clouds). In Target’s case multiple malware alerts were ignored. Some prevention functionalities were turned off by the administrators who were not familiar with the FireEye system. https://arxiv.org/pdf/1701.04940.pdf
And the public cloud is getting more secure all the time. For instance, AWS has been in the process of getting more and more of its services to pass FedRAMP (https://aws.amazon.com/blogs/security/aws-achieves-fedramp-j… ). FedRamp is a compliance standard needed to execute government agency cloud-hosting contracts. AWS is also a PCI DSS 3.2 Level 1 Service Provider, the highest level of assessment available (PCI DSS is generally needed to deal with credit card data, for instance). How many companies can run a private cloud that meets those standards? Some have to, of course.
Note I’m not saying the public cloud is more secure, I’m just saying that it takes a very high degree of knowledge to implement better security in a private cloud than you get for cheap in a public cloud.
As for Latency, even completely in-house applications often need to communicate with remote offices and traveling employees. If one is really interested, this article (https://www.interxion.com/globalassets/_documents/whitepaper… ) is an interesting look at what contributes to application latency, and how it affects some applications more than others. One could even argue that deploying your application in a public cloud enables you to take advantage of geographically distributed servers to reduce latency for a geographically distributed workforce. And Amazon now provides Direct Connect, which establishes a dedicated network connection between your network and one of Amazon’s locations, further reducing latency.
Of course, as the article points out, Today’s modern data centre isn’t just a bunch of rack-mounted servers but a complex web of hypervisors running dozens of virtual machines. This introduces yet another layer of complexity, since the virtualised network infrastructure can introduce its own series of packet delays before any data even leaves the rack itself! . Note that Nutanix run private cloud servers are heavily dependent on virtualization for easy of deployment and management.
It’s interesting that as Cloud Services starting hitting the market, many in the business doubted they would take off, precisely because of the concerns that Denny raises. Yet, what we’ve seen is that the public cloud has steadily gained in popularity despite those headwinds. And even Nutanix doesn’t claim those as advantages, but instead focuses on cost and ease of use. I still believe the trend is towards running in the cloud versus onPrem, but there will always be applications developed for one that should be migrated towards the other. I don’t think Nutanix has a bad business model, but I do think its ultimate popularity is quite limited.