By Patrick Timmons
That which we call a rose, by any other name would smell as sweet
In one of Shakespeare’s most famous plays, Juliet Capulet remarks that a name – in this case, her beloved Romeo’s family name Montague – has no effect on the named’s true nature. The networking industry, presumably coincidentally, has held this belief dear since the inception of the Internet protocol suite in the early 1970s: systems interact with one another using their addresses (absolute), rather than their names. For 2018, I predict the migration for content distribution using addresses to content distribution using names to take its first measurable steps.
The introduction of virtualization en mass into network deployments over the past decade stretched the limits for IP addressing in (somewhat) unanticipatable ways. This is being exacerbated by containerization and the migration of workloads to cloud providers. The data that network consumers are looking for is strewn about networks with abandon: mobile users, migratable servers; a hybrid cloud with its own cloud interconnections; a software-defined edge tiptoeing around a hardware-defined core like a sleeping bear. As technologists, we’re left with the mission to maintain order in an entropic system. We’ve changed our habits, treating configurations and system profiles like code in the inexorable march toward a software-defined everything.
But under it all – under the applications, under your hypervisors – the network remains mostly the same. We’ve innovated on top of the network (and created an entirely new name for that, too: DevOps) to manage and mitigate the problems manifest by IP addressing in today’s networks: address exhaustion and the move from IPv4 to IPv6, private/public boundaries and bridging private networks via a public interconnect, moving virtual machines, etc. However, this innovation largely stops there. Near the bottom of the forgotten IP stack, your machines are still consulting Paul Mockapetris’s white pages (the Domain Name System) to translate names into numbers before launching them into the Ether[net].
The promise of named data networking is that content can be addressed by its name rather than its location, to make easy one of the most prevailing uses of the public internet today: content distribution (CDN). As with most challenges that technologists encounter, there are two basic camps in how we deal with the scalability of modern networks to get at the data we’re seeking: patch/amend/hack existing systems to fix what’s broken or inadequate, or invent something new and different. Generally, but not always, the solutions that get adopted are those that address the most pressing issue(s), while having a trajectory toward the agreed upon “best” solution. I believe we’ll start to see networks reorient themselves toward a naming scheme within the next twelve months.
With any luck, this will start by naming networks and routes to those networks. We’re seeing evidence of this in some of the big, bespoke data centers already. In 2018 these naming techniques will extend outside of a single network’s boundaries to be internetworking techniques.
The loose association of names-to-addresses is about as user friendly as the card catalog at the library of my childhood that supplied the locations to books using the Dewey Decimal System. Let’s aim to reinvent data access.
Patrick Timmons is the Vice President of Customer Success at 128 Technology.
The original post can be found here.