The cloud has revolutionized the way enterprises purchase and access software, the speed at which they develop new applications and how they design and build their network infrastructure. However, despite the incredible hype and near-pervasiveness of cloud in any conversation about enterprise IT, analyst firm – Forrester – claims that only about 30 percent of large North American enterprises have adopted public cloud solutions in their IT infrastructure.
To get a better idea of what those adoption numbers look like, what trends are driving them and what challenges cloud adoption could create within the enterprise, we sat down with Dave Bartoletti, the principal analyst serving infrastructure and operations professionals at Forrester.
Dave Bartoletti has developed, delivered, and marketed emerging technologies for more than 25 years as a software executive at several high-profile technology and financial services leaders. He was at the forefront of the middleware, web, automation, and cloud computing technology waves as both vendor and consumer. He has been an analyst for more than seven years, and he advises Forrester clients on modern service delivery strategies, including virtualization platforms, cloud computing platforms and management tools, and data center automation.
Here is what Dave had to say:
Cloud Sprawl: What is the adoption of cloud like today across enterprises? Do enterprises think “cloud first” for infrastructure, or is there still hesitation when considering moving certain workloads and applications into the cloud?
Dave Bartoletti: First, it really depends on what you mean by “the cloud.” At a high level, you can look at the cloud as either applications or cloud infrastructure platforms.
With cloud applications – or SaaS – there’s crazy adoption levels. Over 65 percent of enterprises in North America have at least one SaaS application, and most of them have more than five. So, SaaS – SalesForce.com, Workday, Office365 and other things like that – is further along in terms of adoption. And a lot of companies consider that as their first foray into cloud. Many more companies start with a SaaS application as the first thing they do in the cloud than start with something like Amazon Web Services.
With infrastructure and platform as a service, on the other hand, adoption is steadily climbing. But – no – we’re certainly not at a place where we’ve even started to enter a world where the average enterprise says, “cloud first,” for their new infrastructure. That’s primarily because there’s such an enormous investment in installed datacenters. But adoption is now taking off.
In Forrester’s cloud adoption forecast, the last 4 or 5 years have been a slow and steady uptick in spending. If you take Amazon Web Services and Microsoft Azure out of the mix, there had been virtually no cloud spending on infrastructure-as-a-Service. That all started to change in 2013 and early 2014.
Today, the percentage of enterprises in North America that have adopted public cloud for data center infrastructure is about 30 percent. That was 12 percent a year and a half ago. That sounds low, but it’s the reality when we look at how many people are moving to public cloud for infrastructure. Now, if you add private cloud into that, it’s a different story. There are 30-40 percent of large enterprises in North America that have some sort of private cloud that is at least in production.
So, it’s a tough question to answer overall – it depends on the type of cloud you’re talking about – and it also depends on what type of buyer you’re asking.
If you analyze existing enterprise companies vs. cloud native, startup or small software venders – the folks that are all within 50 miles of me in the Bay area – you’ll get vastly different results. If you ask people in companies where everyone wears jeans and has MacBook Airs if they’re utilizing the cloud, you’re going to get 100% adoption.
But, to put that in perspective, we are now entering what Forrester calls the “hyper-growth stage.” We’re running up into about 38 percent year-over-year growth in adoption. By 2018, we expect 50 percent of large enterprises to have a production workflow running in the cloud, meaning they’ve actually trusted the public cloud to run a significant portion of their infrastructure.
And what’s guiding that? A huge wave of application analysis.
The cloud has caused every large enterprise to look at its application portfolio and say, “We have 1000 applications we manage. 300 of them, we shouldn’t even be running. Should we be running email servers? No. We should be migrating email servers to the cloud.”
Then they look at another set of applications – like their commercial applications with no SaaS option – and look to the cloud to modernize those applications. And then they look at the applications they built internally and own the IP for. They can use the cloud to make them elastic, scalable and mobile. The greatest cloud motion is occurring in those elastic, customer-facing applications that are a great fit for the economics of cloud.
Cloud Sprawl: What challenges do cloud solutions and resources create for enterprises today that they didn’t experience with traditional enterprise IT infrastructure? What causes these challenges and issues?
Dave Bartoletti: Well, there’s three buckets of challenges. There’s a “fit challenge.” There’s a “business challenge,” and then there’s a “technical challenge.”
The fit challenge involves identifying the benefits an enterprise would get by moving a workload to the cloud in the first place. Enterprises that run traditional IP the classic way have to support an application, figure out what the requirements are for that application, and build the infrastructure that they need to support it. What’s different in cloud infrastructure is that you don’t custom build infrastructure for your workload – you fit your workload to the cloud. It’s a whole different design process.
Amazon doesn’t say, “What do you want? How many CPUs? How many discs?” Instead, they say, “Here’s the ten sizes we have, pick one.” So there’s a fit and a sizing challenge, which is an important challenge to tackle. IT teams don’t have a lot of the skills necessary for right-sizing applications for the right infrastructure. Instead, they simply buy as much as they can – which is why we have so many underutilized datacenter and cloud resources.
The business challenge is caused by the shift from a CAPEX spend model to an OPEX spend model. It’s very easy to predict costs in a datacenter when you work on an annual budgeting cycle. Cloud turns that on its head.
As you start consuming resources on-demand, you need an entirely new cloud business model internally. It upends the relationship technology managers have with the business. Are they now a service provider where the end users and enterprise are just leasing things from them, or are they still a cost center where the enterprise buy stuffs and they run it for them? So that’s the core business challenge.
The core technical challenge includes understanding the difference between how cloud services work and how traditional integrations, databases, and network configurations all work. It all worked in the datacenter, but it’s a fundamentally different thing to understand how to install a load balancer in the datacenter and how to get applications to use Amazon’s elastic load balancer in the cloud. It requires architectural redesign.
Ultimately, the cloud creates technical challenges across all three main things It organizations need to worry about when they run any infrastructure – capacity, availability and performance.
Capacity planning is different in the cloud. Clouds are supposed to be infinitely elastic. So, you don’t do traditional capacity planning because there’s supposed to be as much there as you want. Capacity planning used to be on infrastructure components, and now it needs to be more focused on cost analytics. You’re not trying to determine how many servers to buy anymore, you’re trying to figure out how much more you’re willing to let this group spend on Amazon Web Services.
In terms of availability, how are you going to make sure that things are running when they’re supposed to – because IT should still be responsible for that? It’s a new challenge. How are you going to monitor the cloud? You can’t guarantee availability for what you can’t monitor. It’s the technical challenge of learning what monitoring tools and metrics are available in the cloud. At the highest level, it’s trying to understand how you’re going to monitor and control something you don’t own.
And finally, there’s performance. IT is still responsible for making sure the infrastructure they move to the cloud performs at some SLA that they promised to the business. Well, now, instead of being fully responsible for that SLA down to the floorboards, they have to create an SLA out of the SLA provided by the cloud provider. They’re only going to get as good as the cloud provider will promise. That means you need to think in a different way about how you build your applications for high performance. And – just like availability – you need new tools and techniques for monitoring performance in the infrastructure.
What causes these challenges? Two things fundamentally change when you move to the cloud. One, it’s purchased differently – which you’d think was just a business change, but it’s not. It’s a change to how you manage something when you rent it instead of owning it.
The second thing is that it moves things further away from each other. Every time you build a mobile application in the cloud that connects to a database in your datacenter, you just moved two things further apart from each other and put the Internet between them. That’s going to cause a whole range of latency, performance, security, data sovereignty and other technical challenges that are being fundamentally caused by moving things far away from each other.
This is really no different than what we saw in the move from huge, monolithic mainframe applications to client-server applications. That did the same thing – it split two parts of an application apart. The difference back then, both of those things ran on my corporate network. The cloud moves it into someone else’s datacenter that’s separated from me by a network that I can’t control. It puts it under a bunch of performance, security and data sovereignty restrictions that I don’t control.
Successful companies are the ones that approach it from two angles. What do I still have to manage in my cloud? What will someone manage for me? Forrester calls that handling the uneven handshake. When you move something to the cloud, you’re still responsible for major components of it. If you’re not sure what you’re responsible for, you can’t move it yet.
In part two of our conversation with Dave, we discuss an emerging trends in enterprise IT, cloud management and cloud automation, why it’s gaining traction within enterprise IT departments and things senior IT leadership needs to know when looking to better manage their cloud infrastructure.