This post was published on Hortonworks.com before the merger with Cloudera. Some links, resources, or references may no longer be valid.
Companies are evaluating various combinations of cloud computing models to incorporate into their business infrastructure. They’re driven by the benefits of the elasticity of cloud resources and lower computing costs. In a survey of 450 decision-makers, market advisory firm 451 Research found that two-thirds of companies were planning to migrate at least half of their applications to the cloud in the next two years. The challenge is in determining which half would be better served in terms of complexity, risk, and value by doing so.
The move to cloud technology isn’t an event: it’s a journey, consisting of a set of incremental projects. In each one, companies identify what portion of their application portfolio to move to the cloud before working out how and where to move it. In many cases, companies are creating complex, multicloud infrastructures in long-term projects involving business and technical stakeholders.
The Mechanics of Migration
The cloud migration journey begins with knowing your assets. Understanding which applications you’re running, along with their hardware and software dependencies, is critical to planning your move to cloud technology. An asset or configuration database is a vital tool here, storing configuration details and version numbers for each application to make migration simpler.
During this assessment phase, migration teams can also calculate each application’s impact on business processes and find noncritical ones with fewer dependencies. Migrating isolated applications like these will make early pilot projects easier.
This analysis period will help migration teams choose from several options when migrating each cloud application:
- “Lift and shift”: This option is the simplest. It involves running the same application in a virtual machine environment in the cloud.
- Replacement with SaaS: Some on-premises applications may benefit from simple replacement. Software-as-a-Service (SaaS) options can often provide a better application experience than existing applications.
- Transformation: Some applications may be running on legacy hardware and operating systems that cannot be moved to cloud technology. Other applications may benefit from a complete overhaul, turning them into “cloud native” applications that benefit from features unique to the cloud, such as the ability to run many concurrent instances in containers, decomposition into microservices, and API integrations.
- Wrapping: Some legacy applications may be too mission-critical to move or redevelop. In these cases, an integration layer in front of the application can help to make it accessible for those applications that do move to a cloud environment.
Smart migration teams can also combine these options. For example, they may lift and shift at first, but expand the application with cloud-first capabilities later—451 calls this the “lift, shift, and extend” model.
Based on those decisions, teams must also choose from the different cloud models available. Public cloud is popular, but private cloud—in which companies build their own in-house elastic cloud environments using solutions like OpenStack—can be useful in situations where data is too sensitive to move outside the organization.
Money Management
While making these technical decisions, don’t neglect cost management in the cloud. Cost savings is a common driver for moving to a cloud environment, but there are hidden costs to be aware of. According to Rightscale’s survey of 997 IT decision-makers, managing spending was a top concern among companies that migrated to the cloud. Overall, companies wasted 35 percent of their cloud expenditure last year, the survey found.
In the public cloud, it’s easy to consume the range of services on offer without tracking usage, especially if the implementation team gives unfettered access to users. This makes cloud “sticker shock” a common phenomenon. Similarly, while public cloud environments are happy to allow data to migrate in, getting data out of a public cloud environment often comes with a cost penalty. Be aware of this cloud vendor lock-in and check contracts before signing.
The price of transforming rather than simply moving application software can also drive up the overall cost of a project. Business analysis and development skills are expensive.
Governance and Optimization
These issues highlight the need for an often-overlooked aspect of cloud migration: governance. Administration teams must be able to govern cloud usage to ensure that resources are used appropriately and that costs don’t grow excessive.
Self-service portals are a common feature in cloud implementations, but companies may want to introduce a manual review process to approve requests for compute and storage resources, especially in the early days of a computing project. Integrating this provisioning as part of an IT services management system can be especially useful to keep everything under control.
Optimization goes hand in hand with governance in a cloud environment. By monitoring usage and system performance, implementation teams can quickly understand which aspects of the cloud infrastructure and applications layer are meeting expectations.
Set key performance indicators—such as resource usage, costs incurred, and application response times—to ensure that users are getting the service they expect. Have a plan for identifying root causes of any problems and adapting the system to fix them.
Cloud adoption may be a medium- to long-term project, but the advantage of this is that companies can learn valuable lessons along the way, becoming more mature in their migrations to cloud-based services. Done well, it can also set the scene for benefits far beyond simple cost reduction. A well-engineered cloud environment can be a platform for transforming everything from development workflows to customer services.
Learn more about how the cloud can help you manage your data.
Editor's Choice
