The amount of digital data created today is huge, and it’s expected to expand by 2,000 percent by 2020. That’s an enormous amount of information moving through the world’s networks. Perhaps more challenging than the amount of data being stored is the wide variety of information being stored. The rate at which unstructured data is being generated is alarming enough, but it seems like new data types are being created with each passing day.

Indeed, our “digital universe” will grow by 40 percent a year into the next decade, including not only what people do online, especially via mobile, but the information shared between and among connected systems (the Internet of Things). Add to that the impending rise of UltraHD video,and it’s apparent to even the most casual observer that, even when all that data is compressed, the need for bigger and faster networking pipes will soon reach a critical threshold.

What is not always appreciated is just how latency-sensitive all this data really is. Much of the data being accessed via the Web, for example, has to be retrieved via RESTful application programming interfaces (APIs) that need to be able to respond to calls for data access from thousands of devices — simultaneously — in milliseconds.

The initial wave of all that data should be addressed by 10G Ethernet, the price of which is expected to fall sharply in 2015.The technology will be more affordable to more organizations, but access to 10G Ethernet — and the traffic it enables — will almost certainly lead to a need for first 40G and then 100G Ethernet in the data center. Of course, all that data center traffic is then going to spill out to WANs that will by then need access to robust optical networks to handle it all.

Optimizing the flow of data

The network is now at the core of just about every application of any consequence. Managing the interplay of data among thousands of applications requires a level of networking expertise that goes well beyond configuring a LAN. The flow of data needs to be optimized in a north-south direction in the data center, but, in the age of the cloud, east-west optimization is also critical.

The number of IT professionals with the kinds of skills required to perform these feats of optimization and management is limited right now. In the future, policies attached to different flows of data will tell the network how much bandwidth to dynamically allocate. That level of automation can only be achieved when the network itself has the intelligence built into it to recognize that, no matter how fast it may be traveling, not all data is of equal value to the business.

 

Michael Vizard is an independent business writer and the author of this blog. AT&T has sponsored this blog post.