To support the time-sensitive aspect, many have added an intermediate computing node between the IoT device and the central cloud compute and storage node. This can be on premise of a locally-managed service provider.

CAPTURING DIGITAL TRANSFORMATION
Laura LaBerge, a director at McKinsey & Company of capabilities for digital strategy; Kate Smaje, a senior partner; and Rodney Zemmel, a senior partner at the firm, write in their analysis of a survey entitled, ‘Three new mandates for capturing a digital transformation’s full value’: “As organisations continue to navigate an era of massive uncertainty and disruption, digital tech is an increasingly critical differentiator of both strategy and performance. The actions of today’s best-performing companies reflect that fact. For all other companies, three lessons emerge: use digital tech to achieve strategic differentiation on customer engagement and innovation; build proprietary assets, such as software, data and AI, then combine them with a scalable, cloud-based architecture to create a strategic advantage; and focus the quest for digital talent on C-suite and other executives, given the talent integration challenges that many companies continue to face.”

EDGE CONCERNS REMAIN
Part of the digital transformation equation – particularly with the rise of the Internet of Things – is the growing use of edge computing. But while the pandemic has certainly led to a growth in edge and fog computing, as with digital transformation itself, some concerns remain. With edge computing, there is an increased number of attack points open to cyber-attacks. Also, the distance from the IoT devices creates limitations, and so there is a need to put computing power into the edge computing environment because of latency. Nevertheless, much still depends on where the ‘edge’ is, and it very much depends on the communications infrastructure. That aside, with more storage and compute capacity at the edge, organisations can implement AI functionality, further reducing the workload at the central cloud compute node.

With latency being a concern, to what extent does edge computing rely on artificial intelligence and machine learning to boost WAN performance? Well, for example, the pandemic has reshaped the way in which we now consume entertainment. There has been a massive transition from free to air television, cable TV and cinema to streaming services over the internet directly to the home. At present, autonomous vehicles are also nascent, and their foretold prediction is coming to fruition. Whilst streaming services are a big consumer of bandwidth, the individual data volume and speeds are not that great. It’s the new high-volume applications that are causing problems in transmission times both to and from the edge. Even with edge undertaking data validation, there are massive amounts of data involved with each connected and autonomous vehicle.

Traditional methods using compression and deduplication techniques will struggle to move this data between and across clouds – with their inherent latency issues, as well as requiring heavy compute capabilities. The only option that can move this volume of data with low compute overheads is WAN optimisation that uses AI to manage the data flow. However, it isn’t capable of transferring encrypted data, nor does it completely fulfil its promise in terms of improving WAN performance.

WAN ACCELERATION BOOST
WAN Acceleration is, therefore, the answer, and it’s needed where you have a lot of data – including where there is a certain amount of edge computation that needs to go back to a central location. That data includes what is received and transmitted to and from connected and autonomous vehicles, road and smart city infrastructure.

There are two key factors that affect data transmission over long distances – latency and packet loss. Bandwidth is a factor as well, but past 10-15ms of latency. So, adding more bandwidth has little effect on data throughput. WAN Optimisation, which uses dedupe technology, is a great system if you are transmitting data that you have already seen (warm data), and when you have compressible data streams. If the data is “new” (cold data), then the process is typically slow. Every byte sequence has to be inspected. With cloud data that is transmitted over the internet, encryption is a must. If the data is already encrypted then WAN optimisation can only optimise the data if it has the keys – it first has to decrypt, optimise and re encrypt before it can transmit the data. Storage and power means it requires a larger edge node and therefore can increase costs.

WAN optimisation and compression manipulate the data. WAN Acceleration approaches the issue from a different angle: the only way to reduce latency is to move the two nodes physically closer together. So, rather than manipulate the data, it manages the data flow across the connection to mitigate the effects of latency and packet loss by using parallel data streams across the network managed by AI. Utilisation and, therefore, throughput, is 90%+ of the network bandwidth.

As the data is not manipulated in any way, the process is data agnostic and means there is no need to provide keys. WAN Acceleration is data agnostic and can transmit encrypted data. So, the transmission rate is the same as every other data type. Companies looking to deploy digital transformation, edge computing, IoT and the analysis of Big Data could benefit from WAN Acceleration.

HIGHER WAN PERFORMANCE
My top 4 tips for achieving higher WAN performance to enable the innovation for digital transformation, edge computing and IoT as a result of this are:

1. Remember that throwing more bandwidth at latency and packet loss won’t increase WAN performance. This very rarely solves the issue and only increases your cost. If your throughput requirements aren’t high and your data is compressible, think about implementing SD-WANs. Many of these have encryption and WAN optimisation options.

2. Avoid, if possible, solutions that use UDP as the transmission protocol. It tends to be CPU intensive as it is trying to replicate a TCP/IP protocol and tends to have a low bandwidth capability – again increasing the cloud node costs.

3. If your data is uncompressible or encrypted and you have high latency links with packet loss, the WAN Acceleration is an optimum choice.

4. If network performance is key, select your edge and central node capability on network capacity.

Ryan Perera, Vice President, Asia Content & Subsea Networks at Ciena, claims that the world is now operating on a distributed cloud computing environment. Consumer workloads, enterprise workloads, and now even telco’s internal function workloads are being moved to the cloud: public, private, hybrid; then there is also the growth of edge computing.

“Edge computing is an important part of this distributed cloud computing environment…[and it is] a natural extension of the core cloud”, he writes in his article for The Economic Times’ Telecom supplement. He believes there are several key macro trends that will impact on the distributed cloud computing world. They include the advent of 5G, the rise in residential networks demand and the high traffic across data centres.

In addition to his own predictions, it’s important to remember that when there are multiple edge instances, SD-WANs are a great technology. Bridgeworks has proved this with one large, international IT security company, and it has been shown that SD-WANs and WAN Acceleration are very complementary. They can work well together to support digital transformation, IoT, cloud and edge computing. SD-WAN performance is increased with an overlay of WAN Acceleration with solutions, such as PORTrockIT. To summarise, it is not always an either-or case. Technologies can complement each other.

Click here to read the article on Network Computing.