Blog

Three Major Takeaways from GlueCon 2015

By Atos Apprenda Support

glue-conference-2015

A few weeks ago, I attended GlueCon, Eric Norlin’s excellent developer-focused conference in Colorado. Hosted onsite at the Omni Interlocken in Bloomfield, it’s one of the few conferences each year I try not to miss. Beyond the great speakers and the cutting-edge topics, GlueCon is populated by some of the smartest people I’ve ever met. While the sessions pack great value, the biggest benefit I see is interacting with some very smart and innovative people.

GlueCon has a history of being way ahead of the market. Topics that are hot at GlueCon tend to hit mainstream adoption two to three years later. For example, the first time I heard about Docker was when Solomon Hykes spoke two years ago.

With that in mind, here are three big takeaways from this year’s conference.

1. Microservices are the next big architectural movement

GlueCon had more actual microservices practitioners per square foot than I’ve ever seen in a single place. My unscientific guess is that roughly 50% of the keynotes and 40% of the breakout sessions explored microservices-related topics. The talks generally carried a level of nuance around microservices I hadn’t really seen before. For example, James Urquhart of Dell acknowledged the need for data microservices, which was a nice departure from the always-stateless mantra. And while there was recognition that microservices increase certain kinds of complexity, Adrian Cockcroft of Battery Ventures suggested that they provide a benefit of making interconnections explicit — thereby reducing overall potential complexity by an order of magnitude or more. Microservices were definitely hot, and that led to many discussions on how to manage them.

2. Robust monitoring solutions are a must-have

If you’re going to have a totally service based architecture, monitoring and managing those services is an absolute requirement. Whereas monolithic applications can make do with relatively small, self-contained logging and monitoring footprints, a microservices architecture tends toward a high throughput, distributed monitoring solution. I didn’t attend many talks about specific solutions to this problem, but the expected players (Splunk, New Relic, AppDynamics) were all on the docket. One solution that was intriguing was SignalFX, which focuses less on how to collect the multitude of data generated by a modern architecture and more on how to do true real-time analysis of that data. It’s clearly one to watch as collection becomes more commoditized and the value moves up the stack to analysis.

3. Expect AWS Lambda (and similar services) to become a much bigger deal

You might recall AWS Lambda was launched last year to much fanfare. The audience and presenters at GlueCon noticed as well and touted its many benefits in several sessions. Lambda is an operating environment where the unit of abstraction is the function, and data flows are wired as a series of event handlers responding to different data triggers. While it might just sound like a classic event-driven architecture, the killer feature here is that there are no polling listeners and no services running idle, waiting to be triggered. When an event comes in, the deployed function is spun up almost instantaneously, executed, and then spun back down nearly as fast. There’s virtually no overhead, and (at least in the public cloud) costs are only accrued during the actual execution.

Lambda (and other applications sure to copy it) provide an interesting approach to writing a data-driven application, especially in a distributed world. It takes the low coupling/high cohesion pros of a microservices architecture and extends them even further, to the point where an “application” really only does a single thing and we’re using composition as the predominant pattern for full process flow assembly.

Are there challenges to this approach? Of course. It’s a different mindset than procedural- or object-oriented programming, so the transition might be tough. But if GlueCon’s history repeats itself, in two years we’re going to be talking a lot more about function-based applications.

Top photo credit: New Relic/Twitter

Atos Apprenda Support