Inspiration: working as a data center operation specialist, I know data centers around the globe are the fastest growing power consumers, and how much power a Data Could consume to keep up with load capacity growth demands and to be highly available. High availability is critical for the services hosted in a DC and that's what justify for most entities to almost ignore the negative impacts to the climate and humans. A big challenge in DC operation, is to keep up with the high availability demands without resulting in higher consumption of fossil fuel-based power utilities. Sustainability, unintentional negative impacts on the client and renewable energy utilities have never been seriously considered as a major factor in the design process and operations of most if not all Data Centers, especially when compared to the high availability demand.
How we built it: As part of my duties, I do annual review of our Operation & Facilities procedures and implement needed changes based on several findings from audits of DC, also to make sure to align with our corporate's and IT department's mission statements, doing our best to contribute to the growth of the company and its responsibility towards the community. We mainly document each procedure and equipment manuals. There would be a process to be followed and strictly implemented, that way operators have a clear written process for any case that they could face. I like to ask myself about these procedures, if a newly hired operator was thrown immediately inside our DC with nothing else than these documents absolutely no training, can he successfully run it and not make no mistakes affecting the availability or security of the DC? If yes, then we have a comprehensive system of actions, steps and process that can be submitted for approval by higher management, then implemented officially.
What it does: I created Operation procedures documents utilizing unconventional solution / method in IT to reduce Electrical Load Consumption, and approach that never been truly considered or seriously tried.
South Africa reached a point of rising power demands where currently consumption is higher than the supply power from utilities companies'. Therefore, load shedding has been used as a successful approach to drive down consumption, compared to the other option to increase production of fossil fuel based electrical power succumbing to the ever-increasing demands. These documents will converge into what I imagine an enterprise portal first, before building the software that'll be used by operators to input the servers that has been confirmed to have a downtown time where the services running on the server is not active and detail the exact start and end downtime for each server. Keeping in mind to group the servers with downtime together in a cabinet, that's how firstly we reduce power consumed by cooling units (CRAC) since the compressor will be running less than before due to the number of servers that are off and not producing heat anymore. Also grouping them allows us to use the software to control remotely accessible PDU's and shuts down its power supply during the downtime input by operators, then restore the power to the PDU's few minutes ahead of expiry of downtime.
Challenges we ran into: Determining how reliable this option of Electrical Power consumption reduction, can it compete in a world of assets that can cause millions of dollars direct loss damage and lost opportunities for each single second downtime.
Of course, it can, because on the hand of the equation we know the negative impacts to our delicate environment and ecosystems! It's very important to seriously consider the constant yearly increase of Electrical Power Demands by Data Centers and discuss sustainability being affected by continuing to keep up with demands by blindly increasing electrical power supply. And I'm confident in my solution to use this method in IT Infrastructure and how much change it can bring one Data Center after another.
Accomplishments that we're proud of: Tackle two sources of power consumption inside DC's (from CRAC's and idle servers shutting down) by utilizing one approach.
What we learned: Data Centers across the world are the fastest growing power consumers by large margins and it has always increased year over year.
The system can start as a set of comprehensive procedures and manuals for load shedding sessions to be done manually, then by the results of this initial phase we can determine changes if needed to be done for the second part of the system (coding a portal webpage or a software) that takes over controls to power switch mechanism in PDU's automatically according to the man-made downtime database input in it.
What's next for Data Centers, the giants of power Consumption: I'll absolutely consider pitching my solution to the Sustainability Department in my company, which if I have been given a trial and showcased successful goals reaching method, I'm absolutely sure it'll have big impact on IT infrastructure since it's second largest Petrochemical Company in the world and considered as an example that thousands of companies and businesses follow and do the same. That's an impact that could travel far and wide by showcasing it to motivate companies in the IT infrastructure sector.
Built With
- enterprice-networking
- http
- intranet
Log in or sign up for Devpost to join the conversation.