There has been a constant hype about 5G for its gigabit-class bandwidth, low-latency and high-availability. Being essential factors that unleash unlimited possibilities of 5G and its applications, there is another technology that goes hand-in-hand with 5G. It’s called ‘Edge Computing’. Before we jump into the understanding of the concept, let’s have a peek into the current paradigm of cloud computing and why it’s limited in certain scenarios where high-availability and low-latency is critical.
The paradigm of cloud computing mostly revolves around large centralized servers. In recent years, the usage of cloud services has grown rapidly to get more computing power and storage space. Every time data is generated on a user’s device, it travels to the centralized server for the processing which could be thousands of miles away from the user. This approach becomes a bottleneck for the processes that require intensive computation. Latency is the main issue here. Cloud processing is an issue in mission-critical applications that expect extremely low-latency
Why do we need to decentralize cloud
CPUs and storage devices are becoming extremely compact and powerful while also
getting cheaper. A budget smartphone of today is more powerful than a large computer back in 70s. With more devices in with consumers (smartphones, tablets, laptops etc.), it is becoming extremely viable to offload some computational activity to these devices. This will allow the cloud provider to reduce the need for adding more computational resources. It will make the user experience faster with extremely low-latency as less data is transmitted over the network.
What is Edge Computing?
It is an alternative approach to the cloud environment. It brings data storage and computational capabilities closer to the data source which is considered as ‘edge’ of the network. It allows computing resources to be distributed along the communication path, by
decentralizing the cloud infrastructure. Due to this, most of the user actions are processed in real-time. Network edge can be the user’s device, IoT device, router or CSP’s server infrastructure. Edge computing can be used in scenarios where security requirement is not critical.
Advantages of Edge Computing
Privacy: Avoiding to send raw data to servers for processing and storage
Real-time responsiveness: Useful in applications where low-latency is a crucial factor
Reliability: Capable to work even when not connected to servers
Let’s consider the real scenario where a self-driving car is connected to a cloud server.
This car continuously sends the stream of data to the central server. It would be disastrous if the car has to wait for server’s response for crucial decision making. It is speculation that autonomous vehicles generate 40 TB of data every eight hours of driving. Hence, more time is taken for transmission. Here, real-time and quick decisions are essential. Now imagine that we have installed powerful compute hardware in the car that runs the same navigation and decision-making software from the central server. Here, the car can be considered as the ‘edge’ of the network and the compute hardware will be the edge hardware. This built-in edge computing infrastructure will allow the car to make basic decisions in a fraction of second without dependency on the central server.
Consider a smart surveillance camera that has processing capabilities to process the raw video footage. Rather than sending the raw footage to the central server, the camera sends the selective footage to the server only when certain movement is detected.
OTT Streaming services like Netflix and YouTube create a heavy load on the network infrastructure. Edge computing helps create a smoother user experience via ‘edge caching’ Popular content is cached in the server closer to the end-users for quicker access.
How to achieve Edge Computing?
There are two of the most common ways to achieve edge computing. We can either deploy custom software stack emulating cloud services on user’s edge hardware or we can extend the public cloud to multiple point-of-presence (PoP) locations.
Mobile/Multi-Access Edge Computing (MEC)
Multi-Access Edge Computing or simply Mobile Edge Computing is the existing standard for Edge Computing for LTE networks by European Telecom Standards Institute (ETSI). It allows to have computational/cloud capabilities at the edge of the cellular network. It allows the running of cloud applications and processing in close proximity to the cellular user. MEC allows cellular operators to open their Radio Access Network (RAN) to application developers or content providers. MEC application server can be deployed at eNodeB (Part of LTE network)
Some of the use cases of MEC are: Content caching, connected vehicles, enterprise AR/VR/MR, cloud gaming, drone controlling
It is another standard for edge computing coined by Cisco. It often used interchangeably with, but not the same as edge computing. It refers to extending the cloud to the edge of an enterprise’s network. Compute operations are carried out on fog nodes that are closer to the user with wider geographical distribution.
Edge Computing and 5G
Edge computing and 5G often go hand-in-hand. By 2023, 5G will drive around one-fifth of all mobile data traffic, where 25% of the use-cases will depend on edge computing capabilities. A majority of the new 5G revenue is expected to come from enterprise & IoT services, many of which will rely on edge computing. Telcos can gain additional revenue from edge computing. Enterprises can choose to offload their applications that require high-computation and low-latency on 5G operator’s edge infrastructure.
With the use of edge computing, 5G will provide the support of URLLC (Ultra-Reliable Low Latency Communication) to enable use cases such as V2X (Vehicle-to-everything) and Remote-surgery where the end-to-end latency is expected to be in milliseconds.
5G will also provide eMBB (Enhanced Mobile Broadband) capability for use cases that require high bandwidth. Edge computing can be monetized via partner and enterprise applications are hosted on the network edge. The edge computing application environment allows CSPs to host non-telco enterprise applications (workloads) and open the network as a distributed cloud resource.
Security in Edge Computing
As applications are decentralized from the enterprise’s location, it can provide hackers easy access to data due to potential loopholes in network security. The smaller size of edge devices also makes them vulnerable to being stolen or being manipulated.
On a brighter side, wider distribution of data also makes central servers less likely to
cyberattacks and avoid a single point of failure. Edge solution vendor needs to ensure all edge devices and workloads to follow proper security guidelines, encrypting sensitive data. Edge devices need to be patched with security updates from time-to-time.