Dissecting the Data Center: What Can - and Can't - Be Moved to the Cloud

Brought to you by AFCOM

According to the results of a recent survey of IT professionals, 43 percent of organizations estimate half or more of their IT infrastructure will be in the cloud in the next three to five years. The race to the cloud is picking up steam, but all too often companies begin implementing hybrid IT environments without first considering which workloads make the most sense for which environments.

The bottom line is your business’s decision to migrate workloads and/or applications to the cloud should not be arbitrary. So how do you decide what goes where?

The best time to consider migrating to the cloud is when it’s time to re-platform an application. You should not need to over-engineer any application or workload to fit the cloud. If it’s not broken, why move it? For the purposes of this piece, let’s assume your organization is in the process of re-platforming a number of applications and you are now deciding whether to take advantage of the cloud for these applications. There are a few primary considerations you should think through to determine if moving to the cloud or remaining on-premises is best.

Evaluating What Belongs on the Ground or in the Cloud

First, ask yourself: Is our application or workload self-contained or does it have multiple dependencies? Something like the company blog would be considered a self-contained workload that can easily be migrated to the cloud. At the other extreme, an in-house CRM, for example, requires connectivity to your ERP system and other co-dependent systems. Moving this workload to the cloud would introduce more risk in terms of latency and things that could go wrong.

You should also identify whether it is a customer-facing workload or a workload that is primarily accessed via the web. If so, it probably makes more sense to host in the cloud to ensure your end-users are experiencing maximum uptime, performance and availability. Similarly, the cloud is also ideally suited for workloads with variable workload. If you find it difficult to accurately predict the amount of traffic an application will receive at any given moment—and by association, its capacity needs—you’ll benefit from moving that to the cloud and taking advantage of its inherent agility and additional services.

And it may seem obvious, but your organization’s level of experience with governance and security in regards to cloud applications should play an important role in the migration conversation. If this is your organization’s first time working with a public cloud provider and their SLA’s security terms and conditions, you probably don’t want to start out with a core, mission critical application. An experienced team, on the other hand, should feel more comfortable migrating a more complex application and managing the relationship with the cloud provider.

What’s most important to remember is that in today’s on-demand environment, uptime and an acceptable end-user response time are expected no matter where you host applications. Many companies that go to the cloud do so with the expectation that they’re going to save money, but performance, a key enabler of uptime, is closely tied to cost and comes at a premium in the cloud. Without an understanding of each application or workload’s requirements and how they can be met by a specific cloud service provider, you could be in for a surprise when you get the bill.

When It's Not Time to Migrate

Of course, just because certain applications or other workloads might not be ideally suited for the cloud—or vice versa—today doesn’t mean there isn’t room to reevaluate down the road. For example, it could be that your business needs considerable storage capacity and at some point in the future Amazon Web Services announces a new lower cost storage service that makes a shift to the cloud affordable and meets your security and durability requirements for your organization’s archives.

Generally speaking, though, you should consider a shift only when the requirements for an application change. That could mean you’re running out of space and resources on your physical infrastructure, and rather than investing in additional on-premises hardware it may be more economical to take advantage of the cloud’s scalability.

Alternatively, if you’re currently running an application in the cloud and your provider’s SLA changes to such an extent that it’s cheaper to move back to a physical location or the workload becomes more predictable, it might be time to migrate back on-premises. Some cloud native startups have eventually reached a certain critical mass that makes it more strategic and cost effective to move portions of their infrastructure back to an on-premises (or colocation) model.

Thinking Like an IT Pro in a Hybrid World

Despite these considerations, the reality is that the center of technology itself is becoming increasingly hybrid, and we as IT professionals must start thinking about management and monitoring practices in a hybrid context. With that in mind, here are some best practices to help you better manage a hybrid IT environment now or in the future.

Have a hybrid cloud mindset: Despite the cloud’s growing role in traditional data center strategy, on-premises IT infrastructure isn’t going anywhere just yet. To bridge these two approaches to IT—traditional IT infrastructure and cloud services—and prepare for the hybrid IT future, take a workload-centric approach and use cloud, on-premises and a mix of these as best suits each workload.

Embrace DevOps. Independent of the architecture of the application and the logo outside the building where one’s servers reside, a modern application mindset is important. The DevOps movement has brought us new practices, tools and processes that bring benefits to development and IT operations in general. Organizations can adopt the core principles and best practices of DevOps, including an end-user and performance orientation, end-to-end visibility and monitoring and collaboration to achieve a more agile, available, scalable and efficient data center.

Monitor for the hybrid era: Similar to establishing a unified view across on-premises hardware, where your infrastructure might be comprised of any number of disparate vendor solutions, IT professionals must implement a monitoring system that gives them a view across the entire hybrid IT environment. Such a system will allow you to make informed decisions about what workloads belong on-premises or in the cloud. You should be able to see, through a single pane of glass and at any moment in time, when application performance is slowing down or underperforming whether in the cloud or on-premises, and compare relative performance between these two to make informed decisions.

At the end of the day, with so many options about where to host an application – whether it’s on a container, a virtual machine or in the cloud – companies expect both performance certainty and cost efficiency. The best way to meet this requirement is with the proper monitoring tools that provide an understanding of how your applications change over time, and track the actual requirements of that application and its workload.

Build a roadmap: Remember, there is no right way to adopt elements of cloud computing and introduce hybrid IT into your organization; it’s different for every business, and is often a multi-year journey. The best thing for any IT department considering a move to the cloud to do is to build a roadmap. You need to be informed to make smart decisions when it comes to cloud, even if the decision is to do nothing in the immediate future because there is no immediate need for change.

Making such decisions requires developing specific knowledge, including an understanding of how to get the right visibility with hybrid monitoring tools, building processes for migration and testing of apps and economic and capacity planning models that are independent of specific technologies like virtualization or cloud computing.

The key is to build a cloud adoption roadmap based on a workload-by-workload evaluation that considers requirements, workload variability, potential upside, costs and urgency.

This post originally appeared at AFCOM.com