The concept of Edge technology has been around for decades, since the first electronic exchanges were built.
Investment Banks wanting low-latency, fast trading platforms arranged for their server equipment to be co-located next to the Exchange equipment, making Edge technology a desirable asset. Once the trading algorithm had committed the trade, the execution of it was fast, as it had a very short distance to travel.
On this basis, it’s easy to see how Edge technology goes hand in hand with IoT. Much of IoT relies on small bits of information continuously being transmitted and received, the idea of the “Thing” and the processing of it taking as little time as possible.
We’ve seen many reports in the media hailing edge data centres as the answer to latency concerns in the era of IoT, but are they really the holy grail we’ve been waiting for?
The suggestion is that by distributing the processes nearer to the end users, the latency time is reduced. However, the reality is that unless you replicate the huge Big Data stores as well, the processing has to return to its origin point to get vital pieces of information. Ultimately, this will only add to the overall processing time and cause delays in responses back to the customers.
Telephone Exchanges: Also Not the Magic Answer
Several years ago I looked into a BT Telephone Exchange to see if it might be a suitable location for a data centre. From a real-estate perspective it was value for money, but it had some significant shortcomings, making it unsuitable.
Firstly, exchanges provide telephony services to the general public, so are located in dense housing areas. However, dense housing areas are not keen on having diesel generators being tested and potentially running for several days – the noise and fumes are unwelcome.
But that wasn’t the only issue. The old lift system would have needed replacing as it was incapable of holding the weight and size of equipment that needed to be transported up, and the layout of exchanges generally doesn’t support efficient cooling, nor does the height. For the data centres to be diversely connected they would have needed to dig to present other network providers, which would mean ultimately digging up a busy residential street.
So, What Would Work Better?
Before blanketing the country with Edge facilities and disrupting busy neighbourhoods, companies really need to consider deployment. Keeping our trading scenario in mind, deployment probably depends on the application of IoT and whether it is suitable for distributed deployment, or whether it would benefit from being closer to its data processing store.
Each application needs to be presented on its merits and analysed thoroughly. Some IoT applications are high volume, “person-centric” applications. For these applications, it is well worth considering the use of major cloud resources versus dedicated Edge technology. The cost of developing Edge technology’s permanent infrastructure points works against the dynamic nature of the way that IoT works. Many IoT applications get reinvented multiple times, and morph into new ones – making the cloud an option that is more flexible, and one that delivers greater benefits than specific localised supporting infrastructure.
There are other arguments for the use of the cloud. Let’s face it – not all IoT applications will have long lives. IoT products and services are re-inventing themselves at an alarming rate, so companies really need to ask themselves this: What is the point of building dedicated, expensive facilities with a “lifelong” commitment, when you may be able to deploy with the major cloud providers, who may already have an appropriately located resource in place?
Edge technology has certainly got its place, but companies need to be proactive rather than reactive when it comes to IoT, considering the practicalities, rather than the infrastructure or the M&E point of view, before they can be sure they on the road to intelligent deployment.
Simon Johnson is Practice Director, Migrations at Xceed
Image Credit: Chesky / Shutterstock