This is a guest post by Lance Devin, CIO EdgeConneX.
When it comes to alleviating network traffic bottlenecks and delivering proximate, low latency access to content and mission-critical services, edge data centers are the MVP of choice for those seeking a strong competitive advantage. However, like anything worth having, they do present certain operational challenges. Much of this is because edge data centers can come in diverse form factors – there is no one size fits all model of an edge facility. Inevitably, operators and customers alike begin to ask themselves questions including:
– Can one manage a 200kW micro-edge facility in the same way as a 20MW edge data center?
– How can one economically overcome the differences in running hundreds of hyperlocal or Edge data centers versus a few centralized, large data center hubs?
– Is there a need to have a certain number of staff per kW?
– What is the role of remote hands maintenance in all of this?
In arriving at an answer to these questions, it is helpful to consider current events. Much has been made lately around COVID-19 and an increased trend towards lights out edge data center and remote hands management as economies around the world continue through full or partial shutdowns. Certainly, the industry has more than risen to the challenge of dealing with reduced staff while supporting its customers’ needs for added infrastructure capacity to support intensified usage of digital communication, collaboration and entertainment services. But with or without COVID-19, there is a strong business case to be made for data center operators to focus on designing data centers to run lights out by default.
Traditional DCIM is not enough for enhanced scalability
Of course, an early advocation to this argument is noted by 451 Research in a recent report. Despite coining the term Data Center Services Optimization (DCSO) a decade ago, the research firm notes that many data center operators continue to grind in the field of basic data center infrastructure management (DCIM). Whereas DCSO refers to “highly automated and dynamically optimizing management of datacenter infrastructure as a distinct category,” DCIM is much more focused on the basics of “monitoring, change management and alarm handling.”
The report looks at the path to DCSO as encompassing five levels that go from data centers with zero integrated monitoring or controls to those featuring self-optimizing, autonomic platforms that can predict issues and adjust both facility and IT infrastructure. For the data center sector, any widespread move to DCSO would be rooted in a desire to meet customer service requirements at significantly reduced costs. Yet, for many providers somewhere in the middle of those five levels, DCSO is either out of reach or very difficult to achieve in a colocation environment where different businesses have varied needs.
As 451 says, it will take a wholesale “change in expectations as to what constitutes a well-run data center infrastructure.” From my perspective as Chief Information Officer of EdgeConneX, an international edge data center company, pursuing the deployment of lights out data centers is the embodiment of this shift in expectations. Going lights out provides an immediate economic incentive to pivot deeper into the field of DCSO – all with significant benefits to providers and their customers.
Being part of a small data center company that has a global presence has taught me and my entire team to really value two related things: economies of scale and repeatable business automation. One of the challenges with DCIM is that the components needed to operate a data center of any size are typically siloed and don’t automate business processes across data islands. Consequently, it becomes both time intensive and costly to try and integrate a myriad of technology solutions across a global portfolio of data centers. This would be even worse if it were independent solutions for each data center. Imagine, if you will, the customer experience differing site to site? Going with conventional DCIM is, in my opinion, an ineffective way for providers to address constantly evolving operational and customer needs. This is especially true for EdgeConneX where our mission is to deliver whatever edge our customers need – in any market an in any form factor.
Embracing DCSO: Data-driven automation enables unparalleled service and growth
When one is in the business of managing edge data centers ranging between 2MW and 100MW, data and scalability are of vital importance. Early on in my journey with EdgeConneX, we committed to a model of deploying lights out data centers. In the world of edge facilities, the smaller the form factor, the requirement for less staff rises – in some cases, no dedicated onsite staff are needed. This is the fundamental economics of operating on the frontier of the data center industry. Our decision, however, made it necessary to find a cost-effective, powerful way to automate the operations and processes that go into an edge facility’s lifecycle. Thus, was born EdgeOS – a $2.8 million Operating System investment that allows us and our clients to manage, monitor and enable global data center assets, operations and footprint from a single pane of glass.
Most providers unfortunately get hung up on trying to implement DCIM solutions for individual silos – security, communications management, facility management, change management, data and capacity management. With EdgeOS, we go well beyond that, integrating, automating and optimizing all the processes of our datacenters using 18 function-based modules that comprise aspects such as product orders, ticketing, vendor management, energy management, space management, billing and much more. Though still technically an extension of DCIM, it is much closer to the DCSO paradigm of leveraging every piece of data we can in order to deliver data driven transparency and excellence. Data runs the business and this, in turn, enables global scaling.
Here is a quick scenario that simultaneously demonstrates the level of automation and cost effectiveness we generate by employing EdgeOS to manage lights out data centers. Imagine that a Power Distribution Unit (PDU) in a lights out edge facility fails. Most monitoring systems will cascade a series of faults for attached PDUs which in turn overwhelms an operator. With EdgeOS, the operating system itself searches for a root cause, automatically creating an alarm for the operators, and a ticket with maintenance history, and correlated MOPs (“Methods of Procedures” / Change Management) to the PDU vendor without human intervention. This eliminates the cost and time involved in having a dedicated onsite staff member wait for an alarm, decipher the data and decide who to call and when. Additionally, the maintenance vendors have access to historical trended data, alarm histories, ticket histories, and other pertinent information as they form an ecosystem of service to the customer. Whether we need to help a client improve power utilization expenditures across multiple facilities or optimize how we deliver on our Service Level Agreements (SLAs), dynamic, data-driven automation is key.
In a nutshell, EdgeOS represents our step towards the 451 Research conception of DCSO. Covering 1.2 million monitored points in 8 countries across 3 continents, it was created to give us a much-needed advantage in terms of reduced expenditures, enhanced service delivery and acquiring the grail for all data center providers – maximum scalability. Lights out design is the future.
About the author
Lance Devin is Chief Information Officer of EdgeConneX. Mr. Devin has nearly 30 years of experience in executive management, specializing in bringing new products and businesses to market.
DISCLAIMER: Guest posts are submitted content. The views expressed in this blog are that of the author, and don’t necessarily reflect the views of Edge Industry Review (EdgeIR.com).
data center | DCIM | edge data center | edge infrastructure | EdgeConneX | remote management