Edge Computing and Cloudlets
What is Edge Computing?
The IT world of today is being dominated by cloud computing for past decades. It is used for storing, accessing and processing vast amount of data and IT resources outside our computing devices through the internet. But edge computing can be considered as a small-scale cloud computing. Here, an Edge can be considered as the point from which the computing device or the network which contains the device communicates with the internet. The processor inside a computing device or the router or even the ISP can be considered as the edge of the network. So, the important consideration point here is, the edge is situated in near proximity to the IT and IoT devices whereas the cloud servers are situated million miles away from them. Hence, we can deduce from this point that the edge computing is something in which storing, processing and computing happens at the edge of the network. Yes, Edge computing is an open IT architecture which helps in optimization of computing and processing of internet devices and web applications by bringing the computing closer proximity to the data resources. So, edge computing is simply mean ‘Cloud is coming to you’.
Why Edge Computing?
Along with the explosive growth of the computing devices and data resources, a large volume of data being produced. And this is not just for today and tomorrow, but it’ll have an enormous growth day by day. Using cloud computing, we need to send tons and tons of data resources to the cloud server which is millions miles away from the data resources, which incur many issues like bandwidth problems, latency issues, privacy issues and so on. So, rather than transmitting data resources to data centers deployed in clouds million miles away for processing, the edge computing helps an efficient alternative where the data can be processed, analyzed and compute in a proximity point at the edge of the device or network. Only the data which needs a more resource hungry operations and computations can be send to the clouds far away and all the other computing can be done at the edge. Hence, edge computing helps to minimize the long-distance communications in between the data resources and cloud servers, hence reduce latency, bandwidth problems. For example, consider a resource hungry operation, facial recognition where the processing of the algorithm should be done in cloud server which will consume a lot of time, hence, result in latency issue. By using the edge computing, the processing of facial recognition algorithm can be done at the edge of the device or the network, so will result in speed processing and reduce latency and also bandwidth problems. Similar to this, edge computing helps a lot for real time applications like self-driving vehicles where a vast amount of reduction in latency, helps a lot in real time update of the operations and computations.
So, what are Cloudlets?
The cloudlets can be considered as small scale of the clouds, where the main difference is cloudlets are situated in a proximity point to the devices in the network, at the edge of the device network. It is the cloud with in your geographical location. Similar to edge computing, cloudlets help to do the processing and computing for the offloaded process from the device in the network. I need to explain what is meant by ‘Offloading’.
The mobile as well as other IT devices nowadays are being developed embedded with a number of advanced features such as augmented reality, face recognition, natural language processing, gaming, video processing, 3D modeling software etc. These applications usually are resource-hungry, requiring intensive computation and high energy usage. But the mobile devices are resource constrain in terms of processing power and battery life. So, in order to execute these types of applications, the resource intensive applications are uploaded to the cloud using a mechanism called OFFLOADING where all these processing can be carried out in cloud using the resources there, and the results are send back to the IT devices in our hand. Based on the type of tasks and the needed resources, the whole process or a part of the process get offloaded to the cloud for processing.
But as I mentioned above in the edge computing section, sending data from data resources to clouds which are miles away have latency and bandwidth issues. And, if there is a situation where the internet service provider failed to conserve the connection between the device and the cloud server, there’ll be delays, packet loss and interrupt user experience. So, in order to avoid and reduce these problems, the Cloudlet concept was introduced. A standard definition for cloudlet is ‘Cloudlets are mobility-enhanced small-scale cloud data centers that is located at the edge of the Internet’. So, by using cloudlets, the resource intensive tasks can be offloaded to it for processing hence will reduce latency, bandwidth and save a lot of time. Cloudlets’ latency and bandwidth advantages are especially relevant in the context of automobiles, to complement vehicle-to-vehicle approaches being explored for real-time control and accident avoidance. During failures, a cloudlet can serve as a proxy for the cloud and perform its critical services. Upon repair of the failure, actions that were tentatively committed to the cloudlet might need to be propagated to the cloud for reconciliation. Including these, another benefit of using cloudlets are privacy and security conservation. While using cloud for processing, our secure data have to travel to cloud servers miles away, hence security of the data will be in question. Hence, by using cloudlets, all the private data will be processed at the edge of devices and help in the conservation of the security and privacy of data.
Three main features are highlighted in the cloudlet architecture which was modeled by Satyanarayanan[1]:
Soft-state: One of the most important attributes. Once the cloudlet is installed, its entirely self-managing and doesn’t require any professional assistance.
Powerful and well-connected to Internet: It’s a resource-rich computer or a cluster of computers that are well connected to the Internet and available for use by nearby devices. Similarly, cloudlets have efficient and reliable connection to Internet usually through a wired connection.
Available for use by nearby mobile devices: it’s logically near to devices, it means that any mobile devices in the Local Area Network (LAN) has a low latency to the cloudlet and high bandwidth available to transfer data.
According to Satyanarayanan[1], the proximity of cloudlets paved a great help in following ways:
Highly responsive cloud services: Physical proximity cloudlet to an IT device makes it easier to achieve low end-to-end latency and high bandwidth. This is valuable for applications such as AR and virtual reality that offload computation to the cloudlet.
Scalability via edge analytics: The cumulative ingress bandwidth demand into the cloud from a large collection of high-band width IoT sensors, such as video cameras, is considerably lower if the raw data is analyzed on cloudlets. Only the (much smaller) extracted information and metadata must be transmitted to the cloud.
Privacy-policy enforcement: A cloudlet can enforce the privacy policies of its owner prior to release of the data to the cloud by serving as the first point of contact in the infrastructure for IoT sensor data
Masking cloud outages: If a cloud service becomes unavailable due to network failure, cloud failure, or a denial-of-service attack, a fallback service on a nearby cloudlet can temporarily mask the failure.
According to the researches, the Edge computing can get a faster path to success by nurturing the creation of an open cloudlet ecosystem. So, as an overall point both edge computing and cloudlets are being a disruptive technologies for this cloud intensive era, which bring energy-rich high end computing within a single wireless hop of IT devices specially mobile devices, help to reduce latency, bandwidth and privacy related issues, hence draw path to an era with high intensive applications ahead.
References
[1] M. Satyanarayanan, “The Emergence of Edge Computing,” in Computer, vol. 50, no. 1, pp. 30–39, Jan. 2017.
[2] Usman Shaukat, Ejaz Ahmed, Zahid Anwar, Feng Xia, “Cloudlet Deployment in Local Wireless Networks: Motivation, Architectures, Applications, and Open Challenges”, in Journal of Network and Computer Applications, December 2015
[3] IEEE Innovation at Work. 2020. Real-Life Use Cases For Edge Computing — IEEE Innovation At Work. Available at: <https://innovationatwork.ieee.org/real-life-edge-computing-use-cases/>
