Technological advances such as 5G and industrial automation are leading our society into new realms of efficiency. However, it’s also becoming increasingly evident that these advancements are coming at a cost of ecological sustainability. To counter this, any technological advance needs to ensure that it also supports a sustainable approach to its advancement.
Edge computing is no different in this sense. There are two ways to look at the sustainability equation when it comes to edge technology:
- How can intelligence at the edge support sustainability while preventing revenue loss and increasing revenue gains?
- How can data centers collect the increased volumes of data and still process and derive intelligence from that data in a sustainable manner??
Industries that can increase their sustainability by having more sensors, more and better measurements, and better decision-making include farming, energy, material sciences, construction, and mining. These use cases are already quite far along and well-developed.
But we need to ask: How can we sustainably improve and embrace sustainability?
The data deluge issue
According to IDC, data creation and replication is growing at a faster rate than installed storage capacity, only 2% of all data created is being saved and retained, and the amount of data created over the next five years will be greater than twice the amount of data created since the advent of digital storage.
The increasing amount of data available and clearly increasing enterprise-level hunger to use this data will very likely affect the amount of data being stored and hence the ecological footprint of data storage overall.
Powering servers, cooling servers, keeping facilities’ lights (and hence the electricity used to run the operations) on—all of this contributes to a data center’s CO2 footprint. But that’s not the whole picture. You also need to factor in materials (mining and refining), chip fabrication, server assembly, and having a supply chain to move things from point A to B at each stage of the overall process.
Just to give you an idea of what this involves, according to a study conducted by University of Wisconsin and University of British Columbia, a 1-Terabyte (TB) SSD emits about 369 kg of CO2 over a 10-year period. While this SSD’s operating expenditure (OpEx) is only 49 kg of CO2, the capital expenditure (CapEX - i.e., the amount of CO2 generated during its creation) is 320 kg of CO2. Given the speed at which data now needs to be accessed, it’s safe to assume that any high-performance, low-latency computation scenario will require SSDs, and just 3 TB of SSDs emits a little over a ton of CO2.
With data generation estimates now projected to be somewhere in the 180 Zettabyte range (ie,1 billion TB) by 2025, imagine where we’re headed with the SSD requirement for storing all of this data (180,000,000,000 x 320 = 57.6 trillion kg of CO2!!).
How to use edge data efficiently
So how can edge computing arm enterprises to use their data efficiently and reduce their storage spend, either in their own data centers or in some hyperscalers’ cloud infrastructure?
The simple answer: by applying localized intelligence for decisions and actions.
Most business processes rely on collecting data and then using that data when needed within their systems and processes. Either they haven’t investigated edge computing at all or they aren’t using it to its fullest extent because they’re only using their edge data centers for “data thinning” (ie, reducing the amount of data being sent to the cloud).
Enterprises first need to look at how they’re running their businesses and conduct a business process rationalization effort to determine which processes can leverage the economies of scale provided by a central data center or cloud and which processes can enhance their revenues and security by moving to the edge.
The next step is to establish a hand-off relationship between the edge and the cloud: after the edge decisions and actions are made, should these processes send the raw data or just the data of interest to the cloud? Think of this as a feedback loop between real-time decisions and actions based on business rules that get enriched by machine learning activities that produce new and improved business rules which get reincorporated into the decision making at the edge. This enables the conversion of data into digestible, locally usable formats to drive quicker, more powerful, and more relevant business decisions and actions. One can think of this as “data dehydration”, and the data can be “rehydrated” as needed and only if needed.
Enabling real-time intelligence
Real-time intelligence can make edge computing sustainable and also consolidate the number of technologies required to process the data and make decisions at the edge, which can significantly lower costs.
Typically, the four key capabilities needed for real-time decision intelligence are:
- Storage
- Processing
- Aggregation
- Application of business rules
But here’s the key. Instead of using different technologies for each of these needs and creating a never-ending quagmire of stack complication, use a unified platform that consolidates these four key capabilities in a meaningful manner, simplifying your stack while decreasing your hardware footprint for edge and improving the accuracy and quality of the decisions and actions taken at the edge.
Because, in the end, the same things that fuel sustainability also fuel usability. The smarter we can be about our approaches to new technologies such as edge computing, the more we will get from them. It’s that simple.
