Network Outages Expose an Aging, Vulnerable Internet
None of us is getting any younger, and the same is true of the Internet. Experts who track major outages are warning that the Internet is not aging well, and that the network architecture of a decade ago is not built to handle the growing volumes of data and streaming video.
Add to that growing network security concerns and you could make a good case for overhauling the Internet to provide greater stability, agility and trust, observers say.
The key takeaway, according to industry analysts who track network outages, is that seemingly small issues like a database error or a route leak—defined by the Internet Engineering Task Force as a "propagation of routing announcement(s) beyond their intended scope"—can have major downstream consequences.
According to the network monitoring services vendor ThousandEyes, major Internet and cloud services outages during 2015 resulted from routing issues, configuration and database errors and, increasingly, application issues and outright app failures.
The application issue is particularly troublesome given the wholesale shift to "micro-services" designed primarily to deliver applications across enterprises. While Docker and other proponents of these services have made great strides in building up an application container ecosystem, networks and the aging Internet infrastructure remain a potential bottleneck.
Another networking issue, according to outfits like ThousandEyes, is "unforeseen dependencies" in which corporate applications often run on third-party infrastructure that may be unknown even to developers. Hence, a network failure "trickles down to anything dependent on it," the company notes.
A case in point was a 42-minute outage at Amazon Web Services (NASDAQ: AMZN) on June 30, 2015, the same day there were major fiber cuts in California that may have been part of a coordinated attack. It turned out the AWS cloud outage was unrelated to the fiber cuts and, according to a ThousandEyes analysis, the root cause of was a route leak from a datacenter provider in Boston.
"The route leak affected a wide range of services including consumer Internet sites like Yelp, Netflix and Match; [software-as-a-service vendors] such as HipChat and Jobvite; and financial firms such as Experian and Zions Bank," Nick Kephart, a ThousandEyes network outage analysts explained in a blog post.
Meanwhile, ThousandEyes and other outage trackers warn that the aging Internet infrastructure is sagging under the weight of video streaming by, for example, Netflix binge watchers, and a growing array of interactive enterprise applications.
Even the biggest of the hyper-scalers are not immune to network failures. Two of the biggest outages in 2015 were experienced by Facebook (NASDAQ: FB), which crashed for about an hour last January due to network failures along with configuration and application issues.
The social media giant also encountered a series of application issues and internal network errors during a string of outages last September.
A starker example of aging network infrastructure occurred last summer when the New York Stock Exchange, United Airlines (NYSE: UAL) and the Wall Street Journal all experience network outages on July 8. The stock exchange was down for about four hours while the airline's reservation system went offline for 90 minutes, disrupting air traffic across the country.
While large network vendors are betting billions on a new-fangled Internet of Things, perhaps they should first take a closer look at rebuilding and securing the existing Internet before the next major outage strikes.