Mobile devices, wearables, cameras and many other connected devices or better-called “devices on edge” in different organizations and enterprises, generate a huge amount of decentralized data. Moving this data to the cloud to derive various insights and perform further analytics on them seems a very good option, but there is also a huge dependency that adds to the headache if we have everything on the cloud.

Why not to put everything on the cloud

• Continuous consolidation and synchronizing of data on the cloud can drain resources.
• Maintaining a consistent connection to the cloud gets difficult.
• Costs attached to data transfer and data storage on cloud grow significantly over time.
• Delay-induced due to data transfer and processing on cloud put a restriction in providing near real-time analysis.

With the growing Internet of Things, Cisco predicts that the data generated by IoT will reach 850 ZB per year by 2021. Imagine the load the cloud infrastructure would have to undergo to transfer this data to storage and processing servers. The growth of cloud analytics is exponential in the years to come; hence, the need for edge analytics cannot be ignored.

So can edge analytics replace cloud analytics completely?

Honestly, edge analytics is not here to replace cloud analytics completely, but it is here to complement cloud analytics by driving near real-time analytics as it is close to the data source. Let us see how edge analytics empower cloud analytics.

According to the market research firm IDC, around 45 percent of data will be stored, managed, analyzed and kept right where it was produced, at the edge. So, organizations with 100% total cost of operations on the cloud can leverage edge computing to reduce it to 60%.

• Cloud operations cost can be reduced by using a distributed edge computing architecture, where edge devices together process a critical operation, which a cloud device cannot process on its own, thereby reducing cloud dependency.
• A combined edge-to-cloud architecture is critical for any industrial success. For this, experts need to differentiate and define the real-time analytics to be run at various levels, including edge sensor, infrastructure machine, gateway, controller within on premise appliances and racks or in the cloud.
• We are seeing a tremendous growth in sensor technology. By combining the innovations of sensor technology with the reducing hardware costs, we can establish an edge-to-cloud paradigm. Sensors with processing units can help take critical actions in an inconsistent cloud environment and can later synchronize with the cloud. The required architecture can vary as per industry.

So overall, a well-defined edge-to-cloud architecture as per domain and data would be accelerating cloud computing.

How edge analytics work for all industries

Edge analytics benefit organizations where data insights are needed at the edge. Manufacturing, retail, smart cities, energy, utilities, transportation, and logistics segments are leading the way in deploying edge analytics.

Let us look at sectors that can benefit from edge computing & analytics:

1) Retail:

Brick-and-mortar stores are rich with edge devices such as cameras, beacons, sensors, Wi-Fi networks etc. They are looking for competitive advantages that can help them beat eCommerce businesses, and real-time edge analytics can provide them just that. With edge analytics, sales data, images, coupons used, traffic patterns, and videos are created to provide unprecedented insights into consumer behavior. They have perfect infrastructure and devices to explore edge analytics. Moreover, the mobile devices of customers and data generated by store apps, make this number swell more.

Real-time insights are of prime importance since retail stores need to know their customers’ needs immediately when they enter the store to keep them in store. A recommendation or an offer coming after the customer has left store can be of no use. Identifying customers’ behavior data is something that requires heavy processing power on the cloud. Leveraging some processing at the edge like tracking items viewed, picked, and bought can be a good idea. Other than that, metadata can be sent to the cloud lake to get recommendations, offers, etc., keeping the entire process near to real-time. A distributed edge computing architecture can boost this up further.

2) Manufacturing:

Manufacturing is an industry that requires analytics and computing at the edge. Take an example, an average offshore oil rig has nearly 30,000 sensors. They measure gas emission, pressure, temperature, etc., continuously. Connecting these to cloud lake and deriving analysis will be too costly and time-consuming. A majority of this data is actually not required for analytics; hardly 1-3% of data is used for analysis after cleaning the data. It can bring tremendous advantageous if these edge devices knew what analysis needs to be performed and what data needs to be sent to the cloud, thus saving ample bandwidth. Embedding computing capability in the form of complex event processing (CPE), edge devices can filter out noisy data and collect only information that is deemed useful. In the absence of cloud, the distributed edge computing can process this data for analysis, take critical actions, and can later notify the cloud about the updates.

Another example is of a smart production line. We know that in a production line, each process is time bound. Every action has to be taken in line with production processes. Hence, it becomes important to derive analysis at the edge. Pointing out manufacturing defects or anomalies, badly printed stickers, packaging, etc., in real-time can be achieved using edge analytics.

3) Healthcare:

Healthcare is another domain where we are seeing a huge surge in the number of connected devices. In the near future, a hospital room on an average will have 15 to 20 medical devices, a majority of which will be networked.  A large hospital can have as many as 85,000 connected medical and IoT devices, putting a massive strain on the cloud network. Edge computing and analytics can reduce this burden to a great extent. Here again, real-time analytics will carry more importance than delayed analytics. For example, a clinician’s mobile device is the edge between the patient who is the data source and the cloud. A clinician treating a patient with a tablet will be able to enter patient data into the analytics platform at the edge where it is processed and displayed in near real-time. Patients no longer need to wait for analytics results, which may reduce their number of visits.

In addition, edge computing in healthcare offers another concept called collaborative edge. In a collaborative edge, geographically dispersed data can be fused by creating virtual shared views. This shared data is exposed to the users through some pre-defined interfaces, which edge devices can directly consume.

To sum this up, with edge computing practitioners and patients can get the best response times from the data that is generated and collected by healthcare facilities. As the healthcare sector is using more and more medical devices that are connected to a common network, edge computing is about to become a standard in health IT infrastructure.

eInfochips provides cloud computing services for multiple industries such as transportation, manufacturing, and retail, covering the most common use cases for surveillance, security, and customer intelligence. We also have expertise in providing edge analytics to help clients in a highly-scalable, reliable, and cost-efficient infrastructure with custom solutions for IoT and AI. To know more, get in touch with us.