The internet is surely the most important development in telecommunications of all time, but the technology of the internet is the result of a whole series of steps driven both by demand and regulations. That will be just as true in the future as it has been up to now.
Looking at demand trends, the future internet infrastructure will be made up of the same elements as it is today, but in a different -- even radical -- mix of fiber deployment, edge hosting and content delivery networks (CDNs). The next transformation will change what operators use to build the infrastructure, how services are delivered and how it will ensure the future internet matches expectations.
First, the foundation of the future internet infrastructure will be a vastly increased use of fiber optic cable, particularly of agile optics. Of all the layers of networking, only the physical-layer fiber network actually reduces operations cost with increased deployment. That's because improved fiber connectivity reduces failures and congestion that higher Open Systems Interconnection layers have to handle. Improved fiber connectivity will be used in the future to simplify the design of those higher layers. Obviously, fiber is in use today, but the future internet will shift Capex decisively in favor of fiber technology.
Router experts know that as much as 85% of router code is dedicated to computing data paths, both in the normal state and in response to faults or congestion. Additional fiber capacity could eliminate congestion, even when it's associated with sudden bursts of traffic, like streaming video. Additional fiber routes could ensure every connection has multiple alternative paths. And if a good path around a failure could be substituted quickly, network availability would be much better.
In addition, all of this could be done without adding network management costs. In fact, they could be reduced. Suppose we had a resilient optical network that never reported a path failure. We could then eliminate handling those failures at the electrical level, which means eliminating the cost of the devices and their operations. This could cut the cost of network infrastructure management by 60% or more, according to a CIMI Corp. survey of tier-one through tier-three global network operators. Operators are already pursuing this approach to relieve their profit-per-bit pressure, and the future internet will rely on optical agility.
Edge hosting enables event processing
Future internet infrastructure deployment will also include a lot more hosting at the edge of the network. Operators recognize the over-the-top (OTT) players are earning better profits than operators for selling experiences by connecting those experiences. Some operators have already entered the OTT market through acquisitions, but the big changes in the internet's hosting model will come when internet of things (IoT) and event processing open the opportunity to sell services that rely on hosting at the network edge.
Event processing means doing something when a specific trigger condition is recognized. You can view today's online ordering processes as event-processing applications, but IoT introduces events from machine-generated conditions.
Unlike online orders, which can take seconds to fulfill, machine events need immediate handling. That means placing the event-handling processes close to the edge of the network, not somewhere farther away in a data center. Cloud giants like Amazon, Google and Microsoft all see this to-the-edge trend developing, and they are promoting the cloud as the natural front-end hosting point for event handling. Amazon is even supporting the migration of event processes out of the cloud and into a compatible edge device owned by the user.
A current example of edge hosting event processes in the internet today is found in CDNs that serve the majority of our popular video content. To reduce the effect of internet delay and reliability on video streaming, CDNs cache the content close to the user. It's doubtful streaming video would be practical today without CDNs, and the same is likely true for future applications, especially IoT, that require quick analysis and response to events.
IoT and even personalization will create an explosion in edge hosting, and the edge resources created by these applications will then be available to host components of cloud applications, too. As edge hosting increases, so will the use of fiber to mesh edge-hosting points, which will further increase fiber use and reduce network latency.
Future internet infrastructure could cast CDNs as one giant cloud
CDNs contribute the last piece of the radical transformation the internet will undergo in the next decade. People who stream video don't search for their own caches; they simply ask for a video URL, and the CDN translates that to address the best cache location. Process hosting will work the same way: Users will ask for a process or function and receive the IP address of the best local instance of what they want. Function routing lets operators, OTTs and cloud providers compose experiences based on features, then map them to the most efficient feature edge-hosting points.
If we carry this to its extreme, there could be a time when most internet activity is based on cached content and cached processes. If that happens, we will see the internet start to look like a single giant CDN or cloud, with a gateway through which users access things. This gateway would translate their functional requests to links to specific host or cache points. The internet a user knows would really be created by that gateway point, and the giant cloud or CDN might not use IP internally for many new missions.
In the internet today, CDNs create a kind of shadow network -- a network infrastructure that fills caches with video, maps caches to URLs and touches the internet at these mapping points only. Process hosting would extend this model to IoT and probably even to personalization and location-based services. As this happens, does everyone then have a kind of composed internet, one that delivers cached experiences to each user? This would create something similar to the software-defined WAN model of virtual private network services, where the network's underlying technology doesn't matter.
It won't happen instantly and probably not for every service, but the high-value services of the future seem certain to work in this new way, and so the internet will follow the money.
How to buy a CDN service
Network operators' buying trends change vendor missions
Three telecom tech transformation options for future viability