As in any area of technology, a new generation of network technology promises sizeable increases in performance, functionality or cost – and when the stars align just right, possibly even all of these at once. The question is, how much of a difference from the previous generation is required to justify calling a new technology the next generation above its predecessor?
First, the recognition of a generational gap should come from users, rather than the marketing departments of various device manufacturers. If a technology is truly a big leap over its predecessor, after using the technology the user shouldn’t have to be told – it should be readily apparent and hard to ignore. Using the previous generation should feel like a big step backward.
A good example for network technology is Ethernet. During the evolution from 10 Mbps to 10 Gbps Ethernet connections, the thought was that a 10x increase in headline performance was required to justify a new generation of Ethernet to consumers; people wouldn’t buy a mere 2x or 5x performance boost as it wasn’t worth ripping out their existing gear and didn’t compare favourably to other technological changes taking place at the same time.
Now the answer is a little different; 10 Gbps gave rise to 40 Gbps, 100 Gbps, and more recently 25 Gbps, 5 Gbps and 2.5 Gbps standards have emerged as well. These have been driven by limitations in signal processing and propagation, and in vendors seeking to address a specific application with the standard such as 2.5 Gbps aiming to provide wired connectivity to high-capacity 802.11ac MU-MIMO WiFi access points that can exceed the 1 Gbps capacity of common Ethernet connections, but do not require the bandwidth and cost of a 10 Gbps connection.
Do these newer standards represent any less of a generational gap, if they are not all targeted at improving headline speeds over their predecessors?
In my view, no – although they are not dedicated to providing the 10x increase in headline performance that was offered before, they are making big strides in bringing appropriate and cost-effective connectivity to a range of devices and networks that wouldn’t have been able to utilise it with existing 10 Gbps Ethernet connectivity.
Although headline speeds are important with any technological change, functionality and cost are equally (and often more) important today to drive real adoption of the technology by manufacturers and customers. In the Ethernet example, 10 Gbps connectivity has been available for years – but it has required expensive optical modules, switches and cabling impractical for the kinds of WiFi access point applications 2.5 Gbps and 5 Gbps Ethernet are targeted towards.
So, even though 10 Gbps Ethernet has provided more than enough headline speed, its other attributes have made it unsuitable for the increasingly popular application of WiFi access point connectivity. Despite their lower headline speed, the cost and functionality of these newer Ethernet standards makes them qualify in my view as a new generation as these higher speeds will now be available for many more use cases in an affordable fashion.
A new generation of technology may also not completely replace its predecessor. Again, in the past this was more clear-cut; why care about 10 Mbps Ethernet when you’ve got 100 Mbps now? But today, with a more varied and dynamic network landscape, the answer requires more nuance than before. This is more pronounced when specialised networks are brought into the picture, such as wireless networks and in some ways, data centres.
Wireless networks raise an interesting point in terms of generational gap. With their inherent performance variability compared to their wired cousins, the chances are that future wireless standards (such as 802.11ax from the WiFi world) will not deliver large increases in performance for a small number of devices in an indoor setting, as previous generations have been able to demonstrate very well and successfully use to market products incorporating the technology.
As these technologies target ever smaller niches and applications to deliver increases in performance – for instance, 802.11ax introduces 1024 QAM modulation schemes which are likely to benefit only users who are very close to the access point, and improvements for outdoor WiFi and large numbers of concurrently-connected devices – do they still constitute a new generation in the same sense that previous generations did, such as moving from the headline speeds of 54 Mbps of 802.11g to the 300 Mbps of 802.11n, or are they a different beast?
Now more than ever, our technology is at risk of becoming ‘good enough’ – that is, technologies like 802.11ac WiFi provide what is for most people highly satisfactory performance, and future standards do not look to be able to dramatically increase this level of performance for the current use cases that many people have. When this happens, people are more likely than with previous generations to stick with what they have, and ignore the new technology until one comes along that truly benefits them.
Of course, there are those users who new technologies will benefit greatly, even if they are not for everyone. Continuing with 802.11ax, the amount of brainpower involved in creating the standard is truly incredible and will definitely provide performance improvements for many network deployments – it just seems unlikely to be on the same level as previous generations of technology have accomplished, with overall fewer users seeing dramatic performance improvements.
The answer to the question of whether a new technology is really a new generation for the user will become increasingly dependent on who is providing the answer. For example, a stadium owner may very much appreciate the advancements in outdoor performance and improvements for large numbers of concurrently connected client devices. The average home user however may see no real appreciable difference beyond those devices very close to the access point, and be swayed only by the increasingly large marketing numbers on the box at Best Buy, and the new technology will be seen as an expensive extra with no significant benefit.
So, to answer the original question in the title of this blog post, in a manner familiar to everyone who has ever been or dealt with an engineer: it depends.