Member post originally published on Netris’s blog
Public clouds have basically become identical. Most companies are paying way more for cloud infrastructure than they should. In a nutshell, these statements summarize the key takeaways about the state of cloud computing from the Netris roundtable discussion that Kelsey Hightower recently led.
But pithy summaries hardly do justice to the insights that Hightower and other attendees shared during the event – so, for the sake of those who weren’t able to attend live, we’d like to unpack the cloud computing trends that we explored and the important perspective they offer on what’s coming next in the cloud.
Trend 1: The diminishing differences between public cloud platforms
Early in the discussion, Hightower made a statement that might sound blunt, but is hard to argue with: “As of 2024 most of the cloud providers look the same in terms of their core offerings.”
He was referring to the fact that all of the major public cloud platforms – Amazon Web Services, Microsoft Azure and Google Cloud Platform – now all offer more or less identical sets of core services. To be sure, there are differences in how they implement those services; the specific cloud server instance configurations available from Amazon EC2 are not identical to those on Azure Virtual Machines, for example. But setting aside subtle differences, the big public cloud competitors all offer pretty much the same these days.
The consensus among event participants was that this blurring of the lines between public clouds is not a bad thing. On the contrary, it means that businesses can stop fixating on choosing the right public cloud for their needs – and that using multiple clouds at once will become the norm moving forward
One sticking point that Hightower and others noted, however, is that there remain important differences between the native tooling that each public cloud vendor offers. Being an expert in AWS IAM doesn’t necessarily mean you’re qualified to administer IAM on Azure, for example, because each cloud’s IAM framework involves different concepts and tools.
So, finding ways to simplify tool sets and management processes is important for teams that want a seamless experience working across clouds, even though the clouds themselves now all do pretty much the same things.
Trend 2: On-prem and public cloud have blurred together
Hightower made another bold statement when he declared that as of 2024, the differences between on-prem environments and public cloud environments have basically ceased to matter.
Today, he said, businesses should focus “less on cloud vs. on-prem and more on understanding how the two worlds are converging.”
That’s a bold statement because in many ways, of course, there do remain important differences between on-prem and the public cloud. A bare-metal on-prem server gives you a lot more control than you’d get from a server running in the cloud, for example. Each type of environment is also vastly different from perspectives like infrastructure management and security.
But Hightower doesn’t mean that on-prem and public cloud have literally become identical. Instead, he was referring to the fact that modern tooling (like Netris, which brings the simplicity of public cloud network management into any environment, including on-prem) has made it possible to achieve a pretty similar management experience for your workloads whether they run on-prem or in the cloud.
To drive home the point, Hightower also observed, “I hate the term hybrid cloud because it just creates confusion. There is no hybrid anything. There are just data centers.”
Here again, that might sound a bit strange because there are architectural differences between a hybrid cloud (which combines public cloud services with private infrastructure) and other types of cloud platforms. But Hightower didn’t mean that hybrid cloud literally is not a thing. He meant that with modern tools, managing a hybrid cloud environment is not really any different from managing an on-prem or public cloud environment.
The bottom line: Organizations should worry less about who owns the data center that hosts a given workload and more about ensuring they can effectively manage that workload.
Trend 3: Bold ideas for cloud cost management
FinOps, meaning the practice of cloud cost optimization, has been a buzzworthy topic for several years now as more and more businesses look for ways to control cloud spending. Optimizing AWS costs, GCP costs and other cloud spending has long been a priority.
Typically, FinOps advice boils down to strategies like choosing cloud server instances that deliver the best tradeoff between cost and performance. But folks at the Netris roundtable offered some other, bolder takes on how to cut the fat in the cloud.
Hightower went the furthest when he suggested that data egress – the fees that businesses pay to move data out of a cloud environment – should be free. “Egress is killing companies,” he said, adding that cloud providers who want to attract more customers should eliminate their egress fees.
We’re not saying you should ask your cloud provider to make egress free for you. They probably will say no, because so far, egress charges remain standard and widespread. But we like Hightower’s ambitious thinking about ways to make the cloud more fundamentally cost-effective by eliminating some of the charges that cloud customers have long just accepted.
Hightower also made the case that businesses with predictable workloads are best served by keeping them on-prem in many cases – guidance that cuts against the grain of the “move everything to the cloud now!” advice that has dominated the IT industry for the past decade or more. He noted that, while the cloud is useful in cases where you don’t know how much capacity you’re going to need, “most businesses do know how many customers they have to support and what they require” – and for that reason, they don’t stand to gain much from the built-in scalability of the cloud.
We’ll mention, too, that this is another reason why the ability to administer any environment using consistent tooling is so critical today. When your tool set and management processes are the same across all environments, you can place workloads wherever it makes the most sense, without having to invest in different management strategies for each environment you support.
Trend 4: VPCs are becoming the norm
Despite all of the discussion about the blurred lines between cloud platforms and architectures, roundtable attendees did agree that there remains one really important type of distinct environment in the context of the cloud: Virtual Private Clouds, or VPCs.
VPC is a network construct in the cloud that creates an isolated environment – in other words, a virtual private cloud. Every cloud user can create at least one VPC.
In addition to isolation, VPCs include the essential constructs (such as DHCP, IP subnets, routes, NAT, load balancers, access groups, and network access controls) necessary to engineer and manage networks. So when engineers talk about VPCs, they are often referring not just to isolation techniques but also to the suite of essential network constructs that VPCs provide.
VPCs have existed on the major public cloud platforms for years, but Hightower believes it’s high time for for on-prem, edge, and bare metal environments to take advantage of VPC, too. “Getting the world to normalize the concept of a VPC is important so we can have an abstraction for networking – similar to what containers did for compute,” he said. In other words, Hightower thinks that more businesses should take advantage of VPCs to streamline and standardize networking for their public & private cloud workloads.
Netris CEO and co-founder Alex Saroyan added that managing networking for VPCs has become easier than ever thanks to tools like Netris. “We give them a very similar API and user experience to cloud, while on the backend we take care of the detailed network configuration automatically,” he explained.
Hightower agreed, noting that tools like Netris can be used to unify VPC networking in the public cloud with VPC networking in private cloud and other environments. Netris handles the majority of complex network engineering automatically.
In short, we expect to see more and more organizations adopting VPCs as part of their private and public cloud strategies. We also hope they’ll leverage tools like Netris to deliver a consistent networking management experience across all facets of their environments.
Conclusion: Rethinking cloud strategy
The bottom line: Many of the debates that were at the center of cloud computing for years feel increasingly irrelevant. Today, the most important questions for businesses to answer are no longer which public cloud platform to choose or which workloads to keep on-prem. Modern cloud tooling and networking abstractions have made various environments more similar, which means that moving workloads from one environment to another has become more of a business-driven decision and less of an engineering hassle.
Thus, the top priority for cloud strategy in 2024 is to ensure that businesses can manage their workloads effectively, no matter where they host them. Teams should stop worrying about which labels they slap on their cloud architectures or environments, and more about whether they are administering all components of their clouds as efficiently and reliably as possible.