We talk a lot of about telemetry in the networking world, but generally as a set of disconnected things we measure, rather than as an entire system. We also tend to think about what we can measure, rather than what is useful to measure. Dinesh Dutt argues we should be thinking about observability, and how to see the network as a system. Listen in as Dinesh, Tom, And Russ talk about observability, telemetry, and Dinesh’s open source network observability project.
In most areas of life, where the are standards, there is some kind of enforcing agency. For instance, there are water standards, and there is a water department that enforces these standards. There are electrical standards, and there is an entire infrastructure of organizations that make certain the fewest number of people are electrocuted as possible each year. What about Internet standards? Most people are surprised when they realize there is no such thing as a “standards police” in the Internet.
Listen in as George Michaelson, Evyonne Sharp, Tom Ammon, and Russ White discuss the reality of standards enforcement in the Internet ecosystem.
What if you could connect a lot of devices to the Internet—without any kind of firewall or other protection—and observe attackers trying to find their way “in?” What might you learn from such an exercise? One thing you might learn is a lot of attacks seem to originate from within a relatively small group of IP addresses—IP addresses acing badly. Listen in as Leslie Daigle of Thinking Cat and the Techsequences podcast, Tom Ammon, and Russ White discuss just such an experiment and its results.
A couple of weeks ago, I joined Leslie Daigle and Alexa Reid on Techsequences to talk about free speech and the physical platform—does the right to free speech include the right to build and operate physical facilities like printing presses and web hosting? I argue it does. Listen in if you want to hear my argument, and how this relates to situations such as the “takedown” of Parler.
Automation is surely one of the best things to come to the networking world—the ability to consistently apply a set of changes across a wide array of network devices has speed at which network engineers can respond to customer requests, increased the security of the network, and reduced the number of hours required to build and maintain large-scale systems. There are downsides to automation, as well—particularly when operators begin to rely on automation to solve problems that really should be solved someplace else.
In this episode of the Hedge, Andrew Wertkin from Bluecat Networks joins Tom Ammon and Russ White to discuss the naïve reliance on automation.
Bluecat, in cooperation with an outside research consultant, jut finished a survey and study on the lack of communication and divisions between the cloud and networking teams in deployments to support business operations. Dana Iskoldski joins Tom Ammon and Russ White to discuss the findings of their study, and make some suggestions about how we can improve communication between the two teams.
Please find a copy of the study at http://bluecatnetworks.com/hedge.
I often feel like I’m “behind” on what I need to get done. Being a bit metacognitive, however, I often find this feeling is more related to not organizing things well, which means I often feel like I have so much to do “right now” that I just don’t know what to do next—hence “processor thrashing on process scheduler.” Todd Palino joins this episode of the Hedge to talk about the “Getting Things Done” technique (or system) of, well … getting things done.
The network monitoring world is rife with formats for packets being measured—every tool has its own format. What would make things a lot better for network engineers is a standard data representation for packet analysis, no matter what format packets are captured in. Jordan Holland joins Russ White and Tom Ammon on this episode of the Hedge to discuss the problem and nprint, a standard packet analysis format and tools for converting from other formats.
TCP and QUIC are the two primary transport protocols in use on the Internet today—QUIC carries a large part of the HTTP traffic that makes the web work, while TCP carries most everything else that expects reliability. Why can’t we apply the lessons from QUIC to TCP so we can merge these two protocols, unifying Internet transport? TCPLS is just such an attempt at merging the most widely used reliable transport protocols.
The NSFNET followed the CSNET, connecting the campuses of several colleges and supercomputing systems with a 56K core in 1986. The NSFNET was the first large-scale implementation of Internet technologies in a complex environment of many independently operated networks, and forced the Internet community to iron out technical issues arising from the rapidly increasing number of computers and address many practical details of operations, management and conformance. The NSF eventually became the “seed” of the commercialized core of the Internet, playing an outsized role in the current design of routing, transport, and other Internet technologies.