Four critical steps to datacenter transformation (2024)

Transforming datacenters is more than virtualization and automation. Here's how to do double the work on half the infrastructure

A global 2000 Enterprise IT group is caught in a groundswell of chaos. The current economic malaise is forcing a challenge from the business to IT to cut operating expenses by 20 percent or greater while preserving capital ferociously.

All this while the IT team is faced with another reality, the main corporate datacenter has six to 18 months left in terms of shelf life. The datacenter’s power distribution and patch panel design was not built to handle the massive density and cooling power requirements. The sprawl of unstructured data, app servers, Web servers, and now virtual machines is proliferating at a pace that will force a space crunch in a time frame that is counter to the challenge from the business in terms of capital preservation and opex reduction.

[ Cut straight to the key news for technology development and IT management with our once-a-day summary of the top tech news. Subscribe to the InfoWorld Daily newsletter. ]

Sound familiar? Is this a challenge you’re facing?

What does a firm do? The standard playbook is consolidate, virtualize, and automate. This strategy is absolutely critical and is part of a target foundation that must be built. However, it will not solve the challenge of above.

A different approach needs to be taken. The approach starts with a fundamental principle: IT delivery must be “as needed/when needed,” and all things IT are services that should be delivered in a real-time utility model. With this as the fundamental theory, below outlines four key steps for firms to institute and apply in 90-day building blocks to achieve radical results. It’s important to note these steps can be executed in parallel and in a continuous, iterative. and concurrent manner.

Step 1. Identify the most critical business applications and largest-consuming applications in terms of IT resources in the datacenter. These become the primary targets for a demand-driven optimization approach to transforming the datacenter. The rule of thumb is that 30 percent of applications typically consume or create need for 70 percent or greater of the datacenter infrastructure. Apply a decomposition process to these applications one at a time or in group of similar types. Measure and map the workload across the IT supply chain in terms of performance, consumption, and attributes of how the work execution is managed and what infrastructure it actually uses. This linkage approach of demand and supply creates a data-driven, objective view to affect change. By changing how you manage the application (dynamically at runtime, virtualized across all layers — work, information, and infrastructure) and changing what it runs on (standardized and purpose-driven infrastructure that leverages network and computing physics), firms can consistently find they can do twice the work on half the infrastructure. That math alone will drive radical impacts quickly.

Step 2. Institute a discipline immediately (in parallel to step 1) that measures and monitors consumption, performance, and understands the IT supply chain dependencies of every application. If you don’t understand what an end to end application view and dependency looks like, how can you virtualize or optimize? You are more likely to create further problems in terms of user experience and sev 1 incidents than improve things.

Step 3. Standardize your management strategy of the IT supply chain across the datacenter from a top-down perspective. Incorporate building blocks of runtime management and service orchestration with holistic virtualization (it is very important to note the “with” and “holistic”). Firms that implement just infrastructure virtualization without workload and information virtualization will negate the optimization they are trying to achieve, as they will create new bottlenecks for the delivery of IT from the datacenter. Firms that do not implement dynamic runtime management of workloads will not exploit the elasticity of their virtualization efforts. You have to bring demand to supply as it happens, not force demand into a predefined supply model.

The second building block is a purpose-built combination of datacenter footprints that provide optimized physics — from energy draw to heat dissipation to quantity and types of cables required to connect, communicate, and run workloads. A unified footprint of network, compute, storage, appliances, and software runtime that matches the types of workloads the enterprise supports creates a simpler, leaner platform engineering model.

The third building block is the life cycle management of provisioning, repurposing, and reprovisioning standard builds on top of the unified footprints. This automation gives the business the ability to flex the infrastructure to meet the fluctuating demands while exploiting optimized footprints and matching dynamic workload management needs of the business behavior.

Step 4. Operationalize these building blocks into a new delivery paradigm. Processes need to support instantaneous adjustments of IT delivery to accommodate business behavior. Approval processes and standard operating procedures need to accommodate this model. Finally, IT must learn from business intelligence and constantly mine, analyze, and leverage behavior data to proactively predict, tune, and adjust the infrastructure.

By leveraging this approach, firms can rapidly make large quality and quantity impacts with very targeted efforts. This model can be leveraged globally now, since industry-leading firms such as Unisys are bringing such thought leadership and strategy to enterprise IT groups through datacenter transformation solution offerings.