Cutting edge: IT's guide to edge data centers
A comprehensive collection of articles, videos and more, hand-picked by our editors
What defines a market? Is it what the seller sells or what the buyer demands? Obviously, a market is made up of both, and so the network of the future will have to conform to the service evolution perspectives of consumers and businesses.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
If current network services have shortcomings in terms of features, that will drive network evolution toward a way to address them for simple economic reasons. But understanding what the network market wants, and how delivering it would impact network operators' infrastructure and profits can be answered by contrasting two basic truths.
Truth #1: What users really want from services is lower cost. Most proposed "new" services, like elastic bandwidth or hosted security features, benefit the user not by adding features but by reducing costs. The trend toward higher speed bandwidth for consumer broadband is driven in large part by competition among operators, competition that recognizes that the price/performance ratio for services is what matters. Thus, the prime directive of the network of the future is to lower costs.
Truth #2: More than 80% of profitable network services originate and terminate within the same metro area. Video content served from content delivery networks in the same metro area as the consumer, profitable business services connected primarily metro sites, and future opportunities like carrier cloud are focused on serving users from within their own metro areas. Mobile broadband investment is also focused at the metro network level. So, new services will probably drive metro network changes first -- and most.
Lower cost services challenge network market assumptions
Let's look at our first truth, that cost is king for the network of the future. If what users want is cheaper service, it puts even more pressure on operators that have already been trying to drive costs down just to sustain profits in our all-you-can-eat Internet age. Most operators say their cost and revenue per bit curves will cross over each other in 2017 or 2018, and most operators have constrained their capital spending already. This means new technologies like software-defined networking (SDN) and network functions virtualization (NFV) would have to deliver operations savings to be helpful. That's a problem for two reasons: the vagueness of operations and the scope of impact.
Neither SDN nor NFV define a specific operations model. The deployments to date for both technologies are in limited-scope applications that don't test operations efficiency at scale, and we don't know how SDN or NFV management would respond to a natural disaster or other phenomena that generated a large number of concurrent failures. It's hard to gain a lot of C-level executive support for a technology migration aimed at cost reduction that doesn't have proven operations practices or cost points. It's particularly hard given that small, easy, early steps don't generate much in the way of benefits but can still present significant risks.
Then there's the kicker: Changing technologies to create the network of the future may not be the best way to reduce operations expense. In fact, creating a new model for operations and billing support systems and network management systems above the network could generate more savings faster and at a lower cost, at least in the near term. If the market wants cheaper services, then an approach centered on operations and business support systems (OSS/BSS) is more likely to deliver them by that critical 2017/2018 deadline, when cost and revenue per bit are either the same or network costs are higher than operators' revenue per bit.
A new OSS/BSS approach would almost certainly be based on "abstractions" of technology using modern intent-model tools. This would ease the transition to new technology by lowering the risk of changes in operating practices. It would almost certainly reduce the opportunity for vendor differentiation in network equipment, whether SDN and NFV are adopted or not. Software is indeed in the future of the network; perhaps not in the network itself, but on top of it.
The rise of cached content spotlights metro network edge
What's inside the network of the future, based on the pressure of the market, really means what's inside the metro network. At the metro level, networks have been a combination of computer technology and network technology for more than a decade, and market pressure is going to change that balance even further toward reliance on servers and storage near the network edge, not the use of expensive equipment at the core.
The primary market influence on metro networking is content delivery. As consumer content appetites grow, both for wireless and wireline consumption, it increases the value of caching the most commonly viewed or most valuable content closer to the network edge. This edge-cache trend reduces the need for capacity deeper in the network and improves the user experience. It can also help differentiate a network operator's own content from over-the-top competition from service providers that don't own their own networks, and it encourages deployment of small data centers close to the network edge.
SDN uses servers to control white-box switches, and NFV uses them to host network features and eliminate the need for custom appliances. These applications would justify greater server/storage capacity near the network edge, since the placement of service control and feature hosting close to the users' connection points would reduce network delay and bandwidth consumption. The more IT resources are deployed in these edge data centers, the better their economy of scale, making them more valuable in reducing network costs -- the first market requirement for the network of the future. These changes will combine to create a metro cloud deployment.
Applications like contextual services mediated by an intelligent-agent process like Siri and the Internet of Things would rely even more on hosting features in a metro cloud and further expand the network edge data center. As service components are interconnected more often, the need for data center interconnect among the edge data centers increases. Gradually, metro networks will begin to look like a ring of edge data centers at the places where wireline and wireless broadband is offered, connected by a lower capacity set of trunks to wide area network service access points. Unlike traditional networks that concentrate traffic from edge to center, the future network would have the most traffic at the edge.
Investment is made where return on investment is easiest to achieve, and that will tend to focus network infrastructure changes at the metro level where fat pipes and cloud data centers offer the best return. The network market of the future, as driven by ROI, is likely to look like a series of highly evolved metro networks joined by low-cost fiber connections, and SDN and NFV will have to support this model to be successful.
SDN and NFV to enable agile metro mobile and video delivery
Carrier content delivery networks linked to content monetization
Architecting the network edge for Internet of Things data