By 2022, IDC estimates that 40% of enterprises will have doubled their IT spending in remote locations complementing their infrastructure in core data centres and cloud. This huge shift in computing to the edge represents tremendous opportunities for businesses along with some potential challenges.
Uncertainty of how to plan, design and implement edge computing solutions is common, and it’s crucial that organisations have a clear understanding of their own requirements and any hurdles that may arise during deployment. While the advantages associated with it are massive, the edge computing model does face obstacles that businesses must avoid.
Promises of edge computing
The emergence of edge computing has been steadily rising in importance and maturity. In fact, Forrester predicts that the edge cloud service market will grow by at least 50% by the time the year is out.
Its explosive growth is driven by its ability to process data closer to the source, so the information that is collected and distributed will travel shorter distances compared to data stored in the cloud. Like any entity, data takes time to travel; the larger the amount of data, the longer it takes to arrive at its destination. Hybrid cloud solutions help optimize the management and processing speed of high volumes of data, which means time is saved and decisions are made at a much quicker pace.
There are different classifications of the edge, such as the cloud or core edge and device edge. In general, edge computing can further help businesses overcome the scalability and network performance challenge. Organisations have traditionally relied upon dedicated, purpose-built data centres. However, the versatility of edge computing allows businesses to align with local data centres to focus on desirable markets without requiring costly infrastructure expansions. This means businesses are able to quickly shift to other markets when economic conditions change and scale effectively to meet demand.
More businesses are viewing data as infrastructure, as they design their IT framework around how they are capturing, managing, and using their data. Organisations are better classifying their data and then deciding what the optimal place for that data is, whether it’s the public cloud, a private cloud, at a colocation site or on premises. As more businesses see data as an asset, collecting data at the edge does come with new challenges and can generate liabilities when handled incorrectly.
Complexity is the enemy
A distributed edge system can be more complex to manage than a centralised cloud architecture. It is good practice to centralise where you can and distribute infrastructure only where you must. Businesses that put too much processing on edge devices soon find that the latency and speed issues they were looking to solve with edge computing come back and, in some instances, make matters worse. Complexity at the edge should be avoided at all costs, as the more complexity you have, the harder it is to govern security, scalability, management and maintenance of edge devices. Complexity can be reduced by leveraging unified control planes and a repeatable consistent edge design.
Edge devices should be purpose-built and only do the minimum of what’s needed to collect, process and transmit data, as well as respond to issues that need immediate attention. Organisations should look for simple solutions that are easy to manage or they will likely need to increase spending on deploying their IT staff to deal with any operational issues that will inevitably occur when edge solutions are overly complex.
Cloud services like AWS Greengrass and Microsoft Azure IoT Edge platforms are examples of device edge services to help simplify IOT use cases and edge architecture. They allow for device registry, device communication, local storage, and synchronization capabilities. The edge computing layer is often designed to locally deliver a subset of public cloud capabilities and can sometimes be seen as an extension of the public cloud.
Security should always top of mind when considering adopting new technologies. One of the biggest security risks businesses face is an increased exposure to attacks due to the manipulation of devices within an edge network. For example, cyber criminals have the capacity to install a bot or backdoor to intercept or divert data. Additionally, with 5G networks expected to become the foundation of many IT applications, the integrity and availability of those networks will become a huge security concern and challenge for many businesses – and rightly so. According to IBM, the average cost of a data breach is $3.9 million USD.
However, even with a larger attach surface, edge networks can be secured by using the correct hardening policies, ensuring the proper threat detection exists and encrypting the data at rest and in transit. Data can be protected on local drives before being moved back to the micro data centre. And by using the proper network segmentation techniques, edge computing can minimise risks by localising any data breaches or cyber-attacks to just one point on the network. In doing this, any affected areas can be isolated without shutting down the entire network. Organisations that benefit from edge computing use the proper network security eliminating single weak points and as a result are much less vulnerable.
Balancing network bandwidth
As more data is stored at the edge and more compute happens remotely, edge computing demands a shift in network bandwidth. Traditionally, businesses have allocated a higher bandwidth to data centres and a lower bandwidth to endpoints. Now, organisations are challenged with balancing more bandwidth across the network when moving IoT devices from the core to the edge.
When deployed successfully, certain data is processed locally without being sent to the cloud so less bandwidth will be required. With the ever-increasing numbers of IoT devices all generating live data, bandwidth savings could be considerable with edge computing. Machine learning models that are trained in the public cloud can then be deployed at the edge for inferencing. Machine learning at the edge which can help offload unnecessary data transfer for use cases like image or facial recognition, however, there is always a delicate balance of sizing your compute resources at the edge versus your network capacity. It is also important to discriminate what type of data to keep or discard at the edge in an effort to manage the storage and transfer of large sets of data.
Solution management at the edge
Across almost every sector, businesses are seeking the benefits of edge computing but continue to face on-going challenges when managing it. No two organisations are the same, and every IT strategy should be unique based on business objectives. Businesses should consider working with a trusted IT partner with the skills and expertise to supervise edge networks and deliver its benefits to support long term business objectives. An effective edge computing model should address network security risks, management complexities, the limitations of latency and bandwidth and maximise the true value of a business’ technology investments.
Whether it is optimising business operations, improving user experience, enhancing existing offerings or pioneering new ones, edge computing promises to affect every aspect of the business. All of this may bring new challenges, but those organisations that are able to harness emerging technology properly will reap the rewards in the long term – while those who ignore developments and resist the shift will ultimately be left struggling to keep pace.