Ripping off the band-aid - My RICON CFP Submission
Well, I did it. I've applied to speak at RICON2015 ( http://ricon.io )
I'm pretty new to the whole public speaking thing, so I figure it's a bit of a long shot, but apparently I like the sound of my own voice, and you can't succeed if you don't try!
Here's the synopsis:
(Feedback / overripe fruit welcomed)
All computing requires data persistence to achieve useful work. Data persistence systems employing eventual consistency with guarantees is believed to be the key to transcending CAP, but doing so on the server alone is a half-measure. In order to transcend the limitations of linearizability and create the next generation data-persistence paradigm, we must explore and evaluate certain foundational beliefs in order to expose deeper truths.
- The notion that there exists a single temporal frame reference across the whole of earth is endemic, and false. Though it may be close enough for human purposes, the reality is that Tokyo experiences a subtly different timeline than does New York. Assuming your data traveled through the center of the earth, these cities are 40ms apart, but that's not just travel time, this is relativity in action; This is concurrency in the truest sense. These cities experience reality itself in different temporal frames of reference, 40ms askew.
- The notion that "Client" and "Server" are fundamentally different types of participants is endemic, and false. In reality, the difference is only one of policy and consistency model; A client is only a less privileged participant in the persistence scheme, which may or may not be permitted to implement business logic.
- The notion that a "Database" is the bounding box of a consistency model is endemic, and false. In reality, A resultset is not just a resultset; It is a fractional replica with a poor consistency model. There is no conceptual difference between a node on the other side of the planet, versus a lower tier of CPU caching; The data consistency model permeates all participants, all the way to the L1 cache of each and every CPU; From big iron, to web browser, to wrist watch; they are all the same.