Data center virtualization is big business—it’s projected to become a $10 billion market by 2023, according to research firm MarketsMedia. The virtues of virtualization certainly are gaining rapid traction as organizations clamor to virtualize servers, storage and network devices as part of a never-ending goal to improve data center performance while cutting costs.
Complete virtualization of all aspects of a data center certainly isn’t a new concept, as respected author Samuel Greengard details in a recent Datamation article on Data Center Virtualization. Around since the 1960s, the ability to virtualize data center hardware has received a welcome boost from cloud computing and the emergence of software capable of elevating data center agility, flexibility and efficiency.
Despite those continued performance improvements, however, data centers remain constrained by power density at the rack level. Since running out of power is never an option, data center architects and operators often over-design, over-provision and over-spend on energy, leaving large amounts of power capacity underutilized or unused altogether.
For the past several years, data center thought leaders have looked at power as the final pillar in achieving Software Defined Data Center status. Similar to software defined approaches used with other data center components, Software Defined Power (SDP) provides a common control plane for managing power components to simplify and improve energy utilization. The value drivers are obvious—increased power density, the release of stranded capacity as well as smarter power awareness and distribution.
The tricky part is driving adoption because data center operators face resource management challenges from every corner. Perhaps a page should be taken from the data center virtualization book, where adoption occurred in three distinct waves.
The first wave of adoption is consolidation, whereby elements of the power infrastructure are abstracted, just as physical parts of a data center were abstracted to create virtual elements. In the realm of power, rack consolidation is a first—and vital—step to reducing costs, avoiding CapEx/OpEx and increasing rack/data center density.
Right now, SDP is just coming out of the first wave to move rapidly into the next, which adds automation to SDP platforms. This next step offers a quantum leap in performance to support power-aware workload orchestration along with infrastructure- and application-aware power management. The ability to leverage AI and machine learning is hugely beneficial at this stage as predictive analytics help anticipate the impact of sudden shifts or surges in power consumption before they affect applications or data center operations.
When it comes to planning, designing and operating high-performance data center infrastructures, SDP empowers architects and managers to automate and manage operations more intelligently while minimizing human intervention and errors.
When SDP moves into mainstream data center operations, the third wave, which encompasses federation, then it gains momentum.
Multi-regional data centers and hyperscale computing environments can take advantage of widespread use of SDP to optimize power management across global sites. Cross-data center orchestration and power capacity planning are streamlined to yield larger reductions in total cost of ownership for servers and other power-hungry data center elements.
The evolution of SDP parallels the journey of virtualization that transformed data centers around the world. The complete transformation took time, but it set the stage for SDDC and myriad productivity, operational and cost benefits. The same will hold true for SDP, especially as the pace quickens to leverage its advantages and put the final pillar in place for the ultimate SDDC.