Managing Workloads and Costs in Complex Data Center Architectures
As Mickey Mouse discovered in the Sorcerer’s Apprentice, a sudden, exponential rise in workflow can quickly become overwhelming. Data centers are being put on alert that the virtually limitless “buckets” of device-generated IoT and IIoTdata funneled into centralized cloud-based networks could literally drown their current architectures.
There’s no doubt that today’s increasingly flung-decentralized data sources have become the Achilles heel of cloud-based systems. Edge computing and the cloud clearly serve different masters but are complementary architectures. Working symbiotically, they can provide highly effective IoT platforms.
Custom-Built Edge-Based Solutions
One suggested data center paradigm is to simply push data handling to the edge of a network. Rather than send data to a cloud server or main data center to be processed, move it closer to the population consuming it. Funneled through a local gateway device, edge-based architectures allow faster access and take much of the pressure off of networks.
Edge-based infrastructures (device, edge, and server) sometimes known as ‘fog’ or grid computing, can be set up to dovetail with IoT and most widely distributed applications.
Because IoT gleans data from multiple sensors, controllers, and connected servers, and across remote locations, processing occurs more ideally at the point of origin instead of in the cloud. With edge-based architecture, the ability to process data near a device allows for near-immediate responses and quick decisions. Not a bad way to go for applications that employ generated data in, say, machine-learned algorithms where autonomous decisions are preferred.
A custom-built edge solution lends itself ideally to the rapid accumulation of real-time data, data that devices themselves can use to keep things moving—without the “long trip” back to the cloud. In these cases, CFOs like edge computing for its cost savings in computing power and bandwidth. CIOs like the speed and accuracy of getting automation data from the source.
Universality of the Cloud
Still a mainstay for big data, the cloud’s universal platform lends itself to third-party and legacy applications. For those that don’t rely on stopwatch-timed responses or localized, device-centric processing power, cloud computing is big data’s efficient workhorse. Cloud-based data centers accept massive amounts of data to a server, data that’s pulled out by customers. This is an ideal structure for videos, music, pictures, large documents, and applications that are not critically sensitive to time. But when it to comes to IoT, where information is needed at the source, we clearly expose the cloud’s “feet of clay”—in the lack of quality, speed and accuracy of data. Trying to square peg a centralized cloud solution into an IoT network can quickly eat up bandwidth and computing resources. Particularly when it comes to architectures designed for IIoT applications where network access and latency can be critical.
Uniting the Best of Both Worlds
Working symbiotically, cloud and edge computing deliver what both worlds need: rapid response and big volume processing. Analytic algorithms can be created in the cloud and subsequently moved to edge device sensors that possess no analytical capabilities. Obviously, in some cases, a construct that embraces the unique capabilities of each architecture is preferred: edge computing for time-sensitive applications and cloud computing to address security and big data needs. A mix of both may be what the future brings, uniting the best of each in terms of cost and efficiency. Data volumes transferred and bandwidth costs will drive the formula that makes the best sense for both CFOs and CIOs.
Michael L. Ross, a data center management consultant who has over 10 years helping large data centers reduce their total cost of ownership notes that the right software management tools can help today’s data centers manage the ideal mix of cloud and edge-based data architectures.
“The numbers and types of edge computing devices are exploding. It is estimated that state, local and higher education markets alone will install more than one million IoT devices over the next three years,” he said. “Network and data loads are going to grow but by how much is unknown. This raises some concerns on how much power, space, potential cooling will be required to handle this new workload.”
The introduction of cloud architecture created a shift in the way we look at cost, design and operation of data centers. Traditional DCIM software was touted as the tool that could provide modeling to assist in handling these issues. As IoT technology becomes pervasive, these same issues are magnified.
One solution is RAMP, a next generation DCIM from Tuangru that solves an array of issues through automation, modeling and soon, AI. This type of software and others like it, are well positioned to deliver insight on how IoT impacts current and future data center costs.
About Jad Jebara
Jad Jebara is president and CEO of Tuangru, a next-generation data center infrastructure management (DCIM) software provider. He previously served as senior vice president of finance and administration at Peer 1 Hosting (now Cogeco Peer 1), a hosting service provider where he was responsible for finance, supply chain, and IT.
About Rajan Sodhi
Rajan Sodhi is CMO of Vancouver, Canada-based Tuangru, a next-generation data center infrastructure management (DCIM) software provider with tools that are as meaningful to the C-suite as they are to operators. Users get actionable intelligence that allows them to reduce IT cost, manage workloads and mitigate outages. Contact Mr. Sodhi at email@example.com.
Tuangru’s next generation data center infrastructure management (DCIM) software is designed for today’s hybrid IT environments. Whether workloads reside on-prem, in edge data centers or in the cloud, Tuangru’s DCIM provides managers with a holistic view of their entire infrastructure for management and optimization. The company was recently recognized as one of the fastest growing companies in North America by Deloitte Technology Fast 500™. Tuangru is also a contributor member of The Green Grid. For more information, please visit www.tuangru.com.