Welcome to the new Dyn Research Blog! We’re certainly glad you’re here, and we hope you like the snazzy new look.
Since the Renesys team joined Dyn in May, the number one question we’ve received is “will you keep publishing the blog?” The answer is yes, absolutely, and we hope to bring you some diverse perspectives on Internet performance from other members of the Dyn technical team as well. Please do let us know what you think of the new Dyn Research Blog, and feel free to suggest topics you’d like us to cover.
A moment for reflection
Looking back over the eight years that we’ve been publishing our observations about Internet structure and operations, I’m struck mostly by how you, our audience, have evolved and grown. In the early days, news about Internet infrastructure appealed to a pretty narrow group of readers within the network operations community. We never had to buy beer at conferences like NANOG, but the rest of the world was more or less content to ignore the dirty details of IPv6, peering and depeering, Net Neutrality, and the evolution of the IP wholesale transit industry.
A few things happened along the way, however, that brought Internet infrastructure into the light of day as a topic that many, many people now care about.
- Mobile access emerged as a powerful worldwide alternative to fixed-line Internet, creating a surge of realtime consumer awareness about problems with the reachability and performance of Internet services.
- With smartphones in their pockets, citizens throughout the Middle East took to the streets in the Arab Spring of 2011. Nobody expected what happened next, least of all the Egyptian government. Internet protest created a fast path to global visibility, and turning off the Internet became the fastest way to create global outrage.
- Newly sensitized to interruptions in national Internet service, a billion people between Suez and Singapore began to realize the fragility of the submarine cables interconnecting Europe and Asia. Devastating cable cut outages, which were all-too-common in 2008, have decreased in severity as providers have responded to consumer and business demand for more reliable Internet services.
- Finally, 2013’s Snowden Affair brought the whole world face-to-face with the realities of infrastructure surveillance, casting a shadow over global enterprise’s migration to the Cloud. Internet operations suddenly became a political issue, as governments grappled with their newfound reliance on international networks and content hosted in distant world datacenters.
At the end of this almost incredible sequence of events, our telecommunications infrastructure (including the Internet) has gone from something invisible, maintained by obscure experts, to a subject of informed debate by the general public. Suddenly, even my mom understands what we study, and why we’re studying it. Yesterday I climbed into the car, turned on my favorite radio station, and listened to an hour of conversation about the merits of device encryption and the implementation of local number portability. Companies with whom we discuss Internet performance are conversant with the list of major Internet exchange point cities, and comfortable with the concept of choosing Internet transit providers in each region who provide reliable low-latency paths to their customers. What a change!
Where do we go from here?
The one thing that’s constant on the Internet is, in fact, change. The Internet of 2015 will probably surprise us. If I had to guess at some of the major themes Dyn Research will be covering in the coming years, though, I’d offer the following broad predictions.
Platform diversity continues to be a rich source of infrastructure challenges. We’ve written before about the ways in which limited hardware and software diversity has contributed to the fragility of the Internet. Software Defined Networking (SDN) efforts will add millions of lines of code to the software ecosystem that keeps traffic flowing; will the attention of many eyes eliminate killer defects? Depending on how quickly SDN emerges from the datacenter, we may be in for interesting transitional times.
Geopolitical effects play a larger role in international Internet connectivity. We’ve examined the ways in which Russia’s takeover of Ukrainian Crimea has been reflected in the underlying Internet infrastructure, a game which is now coming to a head in Sevastopol. More and more, governments are waking up to their role in safeguarding and guiding the development of national Internet, and that’s translated into some significant localized intervention in the way Internet content is hosted and accessed. Will they be satisfied with a mere seat at the table in an idealized multistakeholder governance model? Watch the upcoming ITU plenipotentiary meeting for rhetorical clues, but watch Dyn IP Transit Intelligence for the real story.
IP address space exhaustion becomes the new normal. IPv6 may actually be carrying a healthy fraction of all the bytes on the Internet today, thanks to its adoption by intimately-connected, deep-pocketed large content and large access providers. However, more than 80% of the autonomous systems who participate in the Internet continue to ignore IPv6. IPv4 address space is still cheap and plentiful enough on secondary markets that the IPv4 Internet will probably continue to be the de facto standard for global reachability for at least another decade. Creative people respond to scarcity by learning to do more with less, and we expect to see the IPv4 routing table grow through the consensus injection of smaller and smaller prefixes, well beyond its current size of 512K routes. Knowing human nature, we also expect to see continued growth in the rate of IP address theft through BGP hijacking.
Coordinating global resources becomes a huge logistical challenge. We seem to be spoiled for choice, when it comes to figuring out where and how to host important services. Cloud platforms, CDNs, dedicated hosting, all offer solutions to get branded content and services “closer” to the edge of the Internet, where end users live. With the advent of OpenStack, the barrier to entry for new small cloud providers is dropping fast, and the political pressure from local governments to keep local data local is accelerating that distributive process. Dyn is leading this trend, using a broad array of virtual hosting companies around the world to build our globally distributed platform for Internet performance measurement. The next Internet challenges for enterprises will be logistical — figuring out the best way to direct traffic to users at any given moment, taking into account all the local path problems that can affect end-to-end service delivery. It’s going to take some serious measurement, mapping, and management tools to make sense of this challenge.
It certainly has become more interesting writing this blog, as the audience has grown. The connection between our infrastructure stories and the real world seems to grow closer to the headlines each day, whether that’s an attack on Bitcoin miners, Internet disruptions from air strikes in Syria, impacts to routing and domain services due to censorship in Turkey, or the visible impacts of Apple’s CDN on the iOS8 launch. All these stories are connected now, just as we are all connected by the evolving Internet infrastructure we’ve built together. The Dyn Research team looks forward to exploring it with you!
About the Author
Jim Cowie is the Chief Scientist at Dyn. Previously, Jim was the founder and CTO of Renesys, the Internet Intelligence Authority, which Dyn acquired in 2014.Follow on Twitter More Content by Jim Cowie