Cloud and Edge Computing
Data processing in the computing world has taken several twists and turns since the inception of mainframe systems which served as the work horses for bulk data processing. Along the way, companies have constantly sought to optimize the speed and accuracy at which data is processed while minimizing operating costs by using the latest available technologies at their disposal.
Evolution
In the 1960s and 1970s, mainframe systems were developed as centralized computing systems that terminals were connected to using local networks and remote dial-up connections. Users sent strings of text-based commands from terminals to mainframe systems which processed the data and displayed results on the terminal. Mainframes, till today, are known for high reliability, scalability, compatibility and security.
From the 1980s to the new millennium, the advent of the internet drastically changed the power of data processing and computing in general. Client-server distributed models came to the fore. This architecture partitions tasks or workloads between the centralized providers of a resource or service, called servers (host computers), and service requesters, called clients. The client initiates communication sessions with the centralized server through an internet request and in return, receives content or service. This cooperative model led to the advent of email, network printing, and the worldwide web.
In the early 2000s, mobile phones proliferated society and became mini personal computers that hosted various applications and provided users with information at their literal fingertips. At the same time, wireless internet connectivity began to replace wired networks. Companies (led by Salesforce) began to use the internet (“the cloud”) to deliver software programs or applications to users. The benefits were that anyone with internet connectivity could access and download the software while businesses could remotely purchase the software in an on-demand and cost-effective manner.
Edge Computing Versus Cloud Computing
Cloud computing is the storing and accessing of frequently used computer data on multiple remote servers that can be accessed via the internet (“the cloud”). The model enables on-demand access to a shared pool of computing resources such as servers, databases, networks, storage, analytics, and applications. The three main components that comprise cloud architecture include a front end, back end, and a network. The front end usually features a thin or thick client — a thin client is a computer system that connects to a remote server based environment, where most applications and data are stored while a thick or fat client is a system which can be connected to the server even without the network and does not depend on server’s applications and has it own operating system and applications deployed. The back end is made up of servers — built to store, process, and manage network data — and storage devices. The final piece, the network, could be the internet, intranet, or intercloud.
Edge computing refers to the practice of moving data storage and compute power physically closer to the device or the source where data is generated or most needed e.g., an Internet of Things device or sensor. Unlike cloud computing, information is not processed on “the cloud” through distant remote data centers; instead, the cloud comes to you. The term “edge” is coined from the way compute power is brought to the edge of the network or device to reduce the need for large amounts of data to travel which translates to faster data processing, increased bandwidth, and ensured data sovereignty.
The three main components of edge networks are the device edge, local edge, and the cloud. The device edge refers to the physical location where edge-devices run on-premises. The local edge has two layers — application layer that runs applications that are too complex for edge devices and the network layer that runs network components such as routers and switches. The cloud runs both application and network that manage processing that edges cannot handle.
Advantages of Edge Computing
Just like cloud computing which has numerous organizational benefits — including lower upfront costs for equipment and power, cost flexibility with pay-as-you-go model, and better business continuity and disaster recovery through redundant data backups — edge computing has its own benefits.
Edge computing leads to lower latency in data processing. Data is processed at the device edge eliminating the need for data to travel long distances (e.g., data is sent to centralized server locations in cloud computing) and can accelerate generation of insights. Further, edge computing has lower costs compared to cloud computing as the local area network handles processing meaning less data is sent to the cloud. Edge computing also leads to higher bandwidth which allows data models to be very accurate as real-time response is enabled in data feedback loops. Further, edge computing eliminates the need for internet access as data is processed locally and extends the capability to remote locations. Finally, sensitive data is most secure in this model as data is processed in the location it is collected i.e., inside the local area network. This reduces cybersecurity exposure and maintains data privacy and compliance.
Applications of Edge Computing
Organizations can leverage both cloud and edge computing as needed based on their various needs. Edge computing is better suited for industries dealing with real-time data processing, remote locations with limited internet connectivity, large datasets that might prove too costly to send to the cloud, or highly sensitive data types with strict data governance laws.
For instance, autonomous vehicles which require tons of data from their surroundings to make real-time decisions cannot afford any delays or latency from cloud computing. Similarly, streaming services rely on edge computing via a process known as edge caching in which the most popular streaming content is stored in locations closers to viewers for quick and easy access eliminating lags. Most recently, in medicine, surgeons rely on edge computing for medical robotics to provide real-time data and smart analytics eliminating network reliability issues or bandwidth constraints for such high stakes life-or-death scenarios like surgeries and complex treatments.
Sources
