“Can I completely displace my traditional data center footprint with cloud-based services?” Most CIOs of large enterprises are asking themselves this question (or it’s being asked of them). The cost efficiencies, system-availability improvements and flexibility to scale compute and storage capacity on demand are primary drivers of cloud adoption. But it’s not an initiative to kick off without due diligence. Often, governance structures and fiduciary responsibilities require CIOs to take a business-case-justified approach to this transformation. The benefits must be properly weighed against the costs, both in dollars and risk. This process includes ensuring that the cloud-based approach fits the long-term IT and business strategies and meets the company’s financial, technical, operational and security/risk requirements.
And although “moving to the cloud” has become a ubiquitous moniker, a range of cloud options for data center displacement should be considered. They include evaluating the role that as-a-service offerings (e.g., SaaS, DRaaS and UCaaS) can play—especially infrastructure as a service (IaaS) and platform as a service (PaaS), which are the focus of this article. Public- versus private-cloud considerations are important as well. Public-cloud services are attractive, as they offer low prices thanks to their greater use of shared multitenant infrastructure. Not all applications are suited to the public cloud, however. A hybrid approach using public cloud, hosted private cloud and/or consolidated internal data centers will likely yield the optimal solution. On-premises private clouds can be effective, but they often require a high level of expertise to build and operate cost effectively. Applications that are not cloud appropriate have opportunities to consolidate and potentially replatform to newer technologies. Because each enterprise has unique technical, performance and security requirements as well as risk tolerances, building a business case for the transition initiative is critical.
Here are four steps to ensure the data center transition to cloud-based services is rooted in a justified business case.
Step 1: Fully Characterize the Environment
To ensure data center transition plans are set up for success, the transition team should develop a full characterization of the environment. It includes creating detailed application, equipment and software inventories as well as understanding network connectivity, backup architectures, the disaster-recovery environment and identity-management operations in addition to a current cost baseline (e.g., power, real estate, personnel, equipment and services).
This exercise is foundational to the entire process, as it provides a complete picture of the project scope. In addition, it enables the establishment of an accurate configuration-management database (CMDB), which materially de-risks the transition and enables effective IT operations (think ITIL processes, such as incident, problem and change management). In cases where a CMDB is absent or it hasn’t been maintained and updated (a common occurrence in large enterprises), this task can identify ”orphan” applications and infrastructure that can be eliminated instead of moved, often leading to immediate cost savings.
Conducting the inventory can also uncover software compliance problems that may have been lurking for some time—meaning your software provider may now want a true-up. Determining the licenses owned/used/available and the accompanying contract restrictions can help to avoid licensing conflicts associated with the cloud transition.
Step 2: Map the Applications to Infrastructure
Most enterprises prefer an application-centric approach to transition, as it provides a high degree of control and risk mitigation. Defining an exhaustive application catalog will help identify the potentially displaced infrastructure and application move groups, assist with prioritization on subsequent tasks, and facilitate communications with affected end users.
On the basis of the application catalog, teams should map all applications to both their virtual infrastructure (VMs) and physical infrastructure (e.g., physical servers and storage arrays). Doing so will help determine what portion of the current physical footprint can be eliminated (along with its associated costs) after the move.
Further refinement of the move groups requires application-dependency mapping. The application-to-application interface characterization is necessary to determine which applications will be in each cloud/data center move group. Grouping applications with heavy traffic between them in the same move group is necessary to (1) avoid increased wide-area-network charges, (2) avoid performance issues owing to increased latency between application interfaces and (3) avoid storage/database get/put and throughput charges. Many cloud services charge per transaction for interaction outside the cloud. Looking for these usage patterns, which can easily gut public-cloud business cases, is vital.
Step 3: Identify the Best Cloud Destination
To achieve the optimal cost structure by maximizing the use of public-cloud services, enterprises should prioritize the following approaches:
- Priority 1 = Public cloud
- Priority 2 = Hosted private cloud
- Priority 3 = Consolidated internal data centers
For each application, enterprises should evaluate security and compliance requirements to determine whether the public cloud can meet them. This process requires the enterprise’s IT security function to determine whether the public-cloud solutions meet the enterprise’s security requirements. Sometimes, security and compliance requirements eliminate the public cloud, although improvements in public-cloud security are lessening these concerns for many enterprises.
The next task is to evaluate whether the public-cloud provider’s offering will meet each application’s technical and performance requirements (e.g., specific OEM database and version supported, as well as availability/performance minimums). For each application, the team should model the expected environment (e.g., instances, instance uptime, storage, database gets/puts and throughput) to determine the cost/business case for hosting in the public cloud, including optimizing the model for each billing lever (e.g., determine the amount of time to which a reserved instance can be committed—longer commitments drive lower unit costs). Many cloud-service providers have tools to aid in this modeling, and third-party tools are available also. Some tools have capabilities to collect data from the current environment, automatically compare cloud providers and help determine the optimal product mix for a cloud provider and among hybrid options.
For applications where the public cloud is unviable (for security, technical or performance reasons), enterprises should conduct a similar analysis for private hosted cloud solutions. Applications that prove infeasible for hosted private cloud are candidates for a consolidated internal data center.
Step 4: Determine Candidates for New Technologies and Develop an Overall Plan
For the applications that aren’t moving to the cloud, the team should develop a plan to optimize how they will be architected, implemented and managed in consolidated internal data centers. Replatforming to new technologies, such as hyperconverged infrastructure (HCI), should be a consideration. Although doing so may increase the up-front planning time and cost, a converged or hyperconverged infrastructure will save power, space, cooling and maintenance costs. It may also shrink the transition timeline, eliminate downtime and reduce the overall risk.
Applications that aren’t moving to new technologies will require a tradeoff between cost and risk mitigation. Each asset will require its own due diligence to select the right approach (e.g., lift and shift, warm transfer with swing gear, or equipment refresh).
Once these processes are complete, define an overall transition plan and schedule move groups across potentially all three options (e.g., public cloud, hosted private cloud and internal consolidation/migration) all while balancing ROI versus risk. The plan is likely to suffer constrains from the organization’s capacity/ability to execute and requires coordination with other activities and projects.
Enterprises that are moving their data centers to the cloud are doing so to achieve cost efficiencies, reduce operational issues, focus IT resources on higher-value activities and accelerate business plans. Cloud decisions, however, should involve a fact-based, business-case-justified strategy and implementation plan. Without characterizing the current environment and understanding your applications’ behaviors, the cost-sensitive billable elements of the cloud could end up increasing costs rather than meeting the desired cost and performance outcomes.
About the Authors
Andy Sealock, a managing director for business-transformation and outsourcing-advisory firm Pace Harmon, and Dave Pickens, a Pace Harmon consultant, have a combined 30+ years of IT-optimization and data center experience helping Fortune 500 organizations enhance their IT-delivery models.