Whether your company is navigating a newly remote workforce or using data and AI to uncover new business opportunities, this curated selection of articles provides insight and practical advice.
Not all applications can be run in the public cloud. Here’s how to get the cloud to come to you.
By now, the benefits of the cloud are well known: reduced costs, improved scalability, increased efficiency, and more agility. But not all enterprise applications, workloads, and data are suited for it. Compliance and security issues prevent moving some to the public cloud, while complex dependencies among applications won’t allow others to run there. In some cases, it is financially unfeasible or impractical to move them.
However, on-premises applications can get all of the benefits of the public cloud by moving to an on-premises consumption model. I call this having the cloud come to you. In this model, businesses work with a partner that provides them with enough on-premises infrastructure and services to meet all their needs but you only pay for only what you use. This cloud-like economic model shifts spending from a capital expense model to an operating expense one and helps companies better align their business cycles with their infrastructure spending.
Adopting cloud services for your on-premises environment may sound like a daunting task, but it’s easier than you might think. Take the following five steps to get there.
Step 1: Decide which applications and data should be moved
Start off by determining which applications and data should be moved to the public cloud and which cannot or should not be moved. Put together a comprehensive list of your applications and data, along with their needs and requirements. On what hardware are the applications running? How are they architected? Make sure to include the dependencies for each application, including how they communicate with other applications and how frequently those communications occur. List any regulatory and security requirements for each application and set of data. Factor in seasonality issues, if certain applications will be used more during certain times of the year.
Determine how important each application and dataset is for the business and what IT must do to align business strategy with infrastructure. If some applications and datasets are less important to your company, you can ultimately sacrifice performance when architecting them for the new model. But important ones will need the highest performance possible. So, if you’re a financial firm with high-frequency traders or you need low latency for another reason, be willing to pay more for extra performance.
This is also the time to examine your entire application portfolio and consider retiring duplicate, unnecessary, or redundant applications. A general rule of thumb is that 50 percent of workloads are good candidates for moving to a more optimal environment, 30 percent should stay where they are, and 20 percent should be retired.
Based on all that, you’ll be able to decide which applications should be eliminated, which should stay untouched, and which are candidates for the on-premises consumption model.
Step 2: Determine the business case for making the move
Next, put together the business case for which applications and data should be moved to an on-premises consumption model. Try a technique called ease and impact, which uses a two-by-two matrix: On one axis, rate the ease of moving each application to the on-premises model, and on the other axis, rate each application’s business impact if it moves.
In some instances, it will be quite easy to move an application. For example, you may move an application that runs in a virtual machine on legacy hardware and run it on new hardware on premises. However, in other cases, it may be much more difficult to do—for instance, if the application needs to be rearchitected.
Once you’ve rated every application along both axes, you’ll clearly see which of them are easy to move and will have the most business impact once they’re moved. For each of them, perform financial modeling to compare the costs of leaving the application as is or moving it. First, determine the costs of leaving them as is. So, for an application running on legacy hardware, include how much power it uses, how much cooling it requires, how many square feet it needs in the data center, how much administrators’ time it requires, and how many licenses you need. Add all that up to determine how much it costs to run the application every year on legacy hardware.
Compare that to the costs of moving each application to an on-premises model, based on what you’ll be charged for consumption. That will give you a good estimate of the return on investment for moving each application.
HPEStep 3: Right-size the environment for applications and workloads
After the decision has been made about which applications to move, match the capacity you’ll pay for with your enterprise’s actual needs. Doing that requires right-sizing the environment the applications and workloads are going to be run in.
Let’s say an enterprise has decided to move 50 applications to an on-premises consumption model. It’s impractical to move them all at once. They should instead be migrated in a series of waves. Doing this helps determine the right-size environment. That’s because it can be difficult ahead of time to estimate how much capacity all 50 applications will require. But if you migrate them gradually, you’ll create the new environment in bite-sized increments with each new wave, so that when you finish the migration, you’ll have right-sized the environment.
Step 4: Choose the right tools
So you’ve moved your applications to an on-premises consumption model, but your work isn’t over—in fact, it’s just beginning. Because now you’ll need to manage your applications and environment with the right tools to give end users and IT the most productive as-a-Service experience.
Different job titles will need different types of tools. Developers need tools that let them easily and quickly provision VMs and containers. The best of those tools use automation so that developers can merely click buttons to set things such as machine size, network connectivity, disk size and types, and backup schedules.
IT teams need tools to perform their most important management and troubleshooting tasks, such as monitoring performance and costs, provisioning environments, managing lifecycles, doing backups, and handling compliance issues. And data scientists need tools that let them immediately start using machine learning and artificial intelligence without worrying about provisioning and other back-end tasks.
Step 5: Choose the right partner
To do all this, businesses must choose the right partner that can rapidly provide the infrastructure, services, and tools needed to deliver on-premises cloud services. They should look for a partner that recognizes that most companies live in a hybrid world and that there are no longer one-size-fits-all solutions.
The ideal partner must be comfortable working with on-premises, public cloud, hybrid cloud, and SaaS models, and have experience with a broad set of technologies. Look for a partner that can guide you in finding the right mix of technologies based on your workloads, applications, and business requirements, and has a solid team in place for service and support.
Follow these five steps, and your business can expect to see significant savings and operating improvements by shifting to on-premises cloud services. Companies that make the move correctly can expect to see an approximate 30 percent reduction in CapEx and a 65 percent reduction in the time it takes to deploy projects.
Copyright © 2020 IDG Communications, Inc.