Breakthrough flexibility & performance in third generation of TidalScale software GET THE DETAILS >

The TidalScale Blog

    5 Ways Data Centers are Grossly Inefficient

    Authored by: Gary Smerdon

    Earlier this year, IDC surveyed 301 IT users from medium-sized and large enterprises, asking them questions that allowed the research firm to determine the relative efficiency of those data centers. (For reference, the average data center contained 386 blades and servers, while the largest third of those surveyed averaged 711 blades and servers.)

    The resulting study, “Quantifying Data Center Inefficiency: Making the Case for Composable Infrastructure,” sizes the relative efficiency of various aspects of data center infrastructure operations, including provisioning and processes. Both operational expense (OPEX) and capital expense (CAPEX) costs factored into the ratings.

    If you’re an efficiency wonk, IDC’s findings are grim. When it comes to the routine tasks people repeat every day, typically on a manual basis, data centers are just 55 percent efficient. As for technology efficiency, which factors in overprovisioning to accommodate future workloads and redundancy plans, IDC says today’s data centers are only half as efficient as they could be, with a median efficiency rating of 50 percent. When it comes to the processes themselves – such as the steps required to provision a server – efficiency ranks at a dismal 30 percent. 

    That’s a top line finding. Dig deeper, and you’ll find plenty of room for improvement.

    1. Infrastructure is only 45 percent provisioned. Looking at the total capacity in the enterprise data center vs. what is actually available to either physical or virtual machines running actual applications, most enterprises leave 55 percent of capacity unused.
    2. Resources spend most of their time sitting idle. IDC tallied idle compute hours and unused storage capacity, and found only 45 percent of those resources utilized.
    3. Data centers need to be agile, but they’re not. IDC looks at agility in terms of adding or removing capacity and storage instances as workloads change. Lots of time is lost to removing capacity. In an era of unpredictable workloads and rapid data growth, IDC rates data centers at only 30 percent agile.
    4. Idle resources and overprovisioning result in CAPEX wastage. When deployed traditionally, the inherent inflexibility of data center resources like servers almost doom them to inefficient utilization.
    5. Time spent provisioning resources to meet workloads directly results in OPEX wastage. The more time-consuming the process, the higher the costs.

    The paper goes on to argue the benefits of a composable data center infrastructure – an argument that was ably presented here in a two-part blog by IDC’s Ashish Nadkarni, program director for IDC’s Worldwide Infrastructure Practice. (Be sure to read Nadkarni’s blogs, which make the case for the crucial role of software-defined resources, including Software-Defined Servers, in modern data centers.) Nadkarni also authored IDC’s report on quantifying data center efficiency.

    The fundamental goal behind a composable data center infrastructure is to transform traditionally fixed resources into flexible assets that can be right-sized on the fly – quickly and easily – to fit any workload or data set. IT administrators already do this today with software-defined storage and network solutions.

    But until now, servers have remained a fixed asset. Virtualization solutions have helped get more usability from a single server, but when unpredictable workloads and ever-larger data sets overwhelm existing servers, traditional virtualization platforms offer no help. What enterprises need is a way to marshal the servers they already have in house (or in the cloud) to create one or more virtual systems perfectly sized to handle that task.

    At TidalScale, we have taken that concept and given it a name: the Software-Defined Server. With a Software-Defined Server, enterprises can combine multiple commodity systems into one or more right-sized servers. Users and applications see all the resources associated with those servers – memory, CPUs, storage and network – as belonging to one system. A simple point-and-click interface makes provisioning (and reprovisioning) fast and easy, with users able to configure and boot a new server in as little as a minute. No modifications to operating systems or applications are required. It just works.

    Find out more by viewing the video below. And learn for yourself why some of the world’s most respected manufacturers, financial services firms and research universities are introducing flexibility to their data centers with TidalScale’s Software-Defined Servers.

    Click me 

    Topics: IDC, data center, software-defined data center, composable infrastructure