In previous posts I covered the impact of the Internet of Things (IoT) on the rising volumes of data, what all this data enables in terms of value and the landscape of data management with the place of data lakes.
I briefly mentioned edge computing, a decentralized computing paradigm that gains fast acceptance in data-intensive environments where speed matters and IoT data need real-time analysis for swift actions. In this post let’s look at what edge computing is and can do for you.
Edge computing is one of the hottest topics in the industry for the moment. According to Gartner by 2025 a whopping 75% of data will be processed at the edge (more below).
As the name indicates edge computing brings computing power closer to the edge of the network. This enables to store and analyze data closer to where it is generated which reduces pressure on communication networks (lowering network latency) and the heart of the network (data centers and, increasingly, the cloud).
Edge computing: delivering computing capabilities to the logical extremes
Edge computing also powers emerging applications, is key for mission-critical applications requiring real-time processing and is a solution in case of interrupted/poor access to the network or cloud when this can’t be afforded.
Edge computing is not a replacement of cloud computing. Adoption of the cloud will continue to grow since it is perfect for the processing of large volumes of data and artificial intelligence and machine learning.
In my post on data lakes I also mentioned how data lakes increasingly are moving from on-premises environments to the cloud. Yet, while cloud computing focuses more on big data, with edge computing we are in the sphere of fast, instant data. There are different definitions of edge computing. Under the stewardship of The Linux Foundation there is even an Open Glossary of Edge Computing to define a common lexicon regarding anything related with the edge.
In the latest revision of the glossary edge computing has been defined as “the delivery of computing capabilities to the logical extremes of a network in order to improve the performance, operating cost and reliability of applications and services”.
More important than providing a definition is understanding where edge computing fits best – or what sits better at the edge and what sits better in the cloud.
By way of a reminder: a majority of IoT applications across several use cases today can do with a delay in data transmission and analysis. These are typically use cases which are served by some of the major LPWAN standards with a higher latency as I discussed in previous posts.
Where edge computing fits: use cases and industries
Typical circumstances where edge computing comes in the picture are those that require data to be initially stored and processed locally and where there is no business sense to use cloud computing. In some cases, this can be literally very close to or in the connected ‘thing’.
Connected cars, for example, are among the main areas where edge computing makes sense because there is a lot of data to rapidly process and connected cars, indeed, always need to be connected as we move to autonomous cars. As Gartner points out in a blog with the mentioned forecast, in a vehicle, an edge solution may aggregate local data from traffic signals, GPS devices, other vehicles, proximity sensors and more.
In many other use cases, edge computing needs hardware and software capabilities in the so-called last mile of the network with special edge servers in the field or even micro data centers where the information gets collected and analyzed close to the endpoints. Per Gartner: “Edge servers can form clusters or micro data centers where more computing power is needed locally”.
Edge computing offers tremendous benefits – and often is even a must – in critical environments such as heavy processing, nuclear plants, mission-critical facilities such as hospitals, manufacturing where machines need real-time data to control and optimize processes, healthcare with often very critical applications needing immediate action, security monitoring with video and recognition, and applications in situations with poor connectivity.
However, edge computing certainly isn’t just for critical applications or the industrial use cases of Industry 4.0. Financial services, media and even gaming are among the areas where edge computing investments are on the rise. Finally, as the next generation wireless connectivity standard of 5G is coming and consumers increasingly demand high-bandwidth and real-time applications, edge computing is expected to expand into more use cases than the current ones.
The place of edge computing in a real-time economy
As you probably noticed, the real-time aspect is essential in edge computing. If you want real-time applications (certainly with high data volumes) this translates in a need of low latency (delay) across all elements of the solution and technological architecture powering it.
Since the company I work for, Schneider Electric, is active in several of the mentioned industries and in data centers, edge computing is essential in our offering and roadmap. It, among others, impacts EcoStruxure, our IoT-enabled interoperable architecture and platform that offers edge control and supports edge computing across many applications. Since professional edge computing deployments often require micro data centers as mentioned before it also impacts our data center business.
As you can read here we define edge computing as an IT architecture designed to put applications and data closer to the users or things that need them. Whereas cloud computing drove the creation of few mega data centers, edge computing brings distributed IT with an exponential number of micro data centers.
The infographic below gives a comprehensive overview of the place of edge computing and cloud computing with the analytics happening in the core of the network and at the edge – and where devices/sensors fit in it all. As is the case with any network architecture, edge computing needs servers, gateways and software that supports it (for instance, the IoT platform).
When an IoT edge strategy makes business sense
To conclude an overview/summary of when to choose edge computing (whereby striking the right balance between what sits at the edge versus what sits at the cloud is key as my colleague Cyril Perducat puts it in his post):
An IoT edge strategy makes sense when you need:
- The possibility to make quick decisions by using data as close to its source as possible to avoid network latency;
- To optimize the data flow to the cloud by processing raw data at the edge and only passing the high-value aggregated and contextualized data to the cloud (keeping in mind that in IoT raw data can be leveraged using a data lake as explained in my previous post);
- To have offline access to data analytics when access to the cloud and/or the network is not available;
- A better life cycle management of a large fleet of devices, which can be updated remotely and securely versus centrally.
Top image credit: Altizon