Revue des articles sur les méthodologies de pilotage de projets et de la gouvernance des systèmes d'information.
techniques de gestion de projets IT et informatiques,
stratégie et tactiques pour les SI.
The IT Watcher.
March 07, 2014, 2:00 PM — The adoption of virtualization in recent years has laid the groundwork for many IT organizations to move from on-premise data centers to co-located environments and the cloud, says Craig Wright, principal at IT and outsourcing consultancy Pace Harmon. The increased acceptance of high-density platforms that require much smaller physical locations encourages portability as well.
Cloud implementation continues to grow, whether public cloud for standardized situations or private clouds for solutions that are differentiating or have increased security or regulatory requirements. That's driving more focus on orchestrating and aggregating infrastructure services, Wright says.
And automation is starting to shake things up with the promise of the software-defined data center. "In this scenario, everything in the data center is virtualized -- applications, databases, networks -- and an automation layer extends across all virtualization layers to create a unified platform," says Wright. This emerging approach requires a high level of virtualization maturity and orchestration sophistication to put all the pieces together efficiently.
The benefits of these combined technology advances are clear. As a result, most companies are taking advantage of them to improve data center operations. "Even the doubters are piloting and incubating capabilities," Wright says. But such shifts require IT organizations -- and their outsourcing providers -- to rethink their traditional data center strategies.
Following these six steps will help you stay ahead of data center changes.
1. Invest in skills, training and controls. While automation and standardization may result in reduced overall headcount, they increase the demand for higher-skilled resources. "It's about quality, not quantity," says Wright. Traditional data center professionals managed the environment based on standard operating procedures, predefined event triggers, and alarms. They were hands-on, installing hardware, running cables, and dealing with the physical platforms.
In a heavily virtualized environment, data center professionals must understand logical constructs, such as the dynamic distribution of operating systems across multiple platforms, in order to manage, optimize and trouble shoot. "Deploying qualified resources, investing in continuous education, and aligning service lifecycle controls for internal and external providers should become the norm, not the exception," Wright says.
2. Rethink infrastructure outsourcing. New data center technologies encourage more outsourcing, but at a lower volume than in the past. "In many cases, there is more outsourcing of the infrastructure or platforms themselves, but the control and architectural side more often is being retained," says Wright.
Meanwhile, data center trends are paving the way for increased offshoring. "The labor arbitrage value and benefits are gaining for data center deals," Wright says. "A combination of standardization of and advances in technology is reducing the need for heavy physical environments. Utilizing cloud and virtualization opportunities to execute remote infrastructure and management services, organizations are able to leverage multiple providers in low cost delivery centers."
3. Design for high-density platforms. Old world data centers had an average density or five kilowatts per cabinet. High-density racks more than double that. Bake that into data center plans, because retrofitting an existing data center to manage the power and cooling required can be prohibitively expensive, Wright says.
4. Find the right balance. "Utilize hybrid solutions that play to your organizational strengths," says Wright. "Mixing a small on premise footprint or a colocation facility with cloud services often provides the best of old and new worlds."
5. Factor in legacy applications. A major sticking point is how to deal with legacy applications -- 1980s era technology that's not suited to moving to a new data center environment or which cannot be virtualized. Do you retire and replace them? Keep them running in tandem?
"The reality is that many of these are still stable platforms, and there is no compelling business reason to upgrade," says Wright. "This is a very common issue. How do you keep the legacy platform as a part of the virtualized or automated data center, or what do you do if legacy platforms are anchoring you to an old world data center?"
6. Think like a user. "The biggest challenge is assembling and integrating all these advances into a holistic solution so it looks complete to the end user," Wright says. "Exposing the end user to multiple different user experiences depending on data center location or solutions can be a major cause of business dissatisfaction with IT services." Create a consistent user experience regardless of where a service is delivered from or how it is provisioned.