Velocity 2015: Beyond DevOps, Part 1
Velocity 2015: Beyond DevOps, Part 1
Velocity 2015 is pretty great so far. Here's a bit about responsive images, HTTP2, and getting beyond the DevOps buzztalk to an honest focus on the user.
Join the DZone community and get the full member experience.Join For Free
Download the blueprint that can take a company of any maturity level all the way up to enterprise-scale continuous delivery using a combination of Automic Release Automation, Automic’s 20+ years of business automation experience, and the proven tools and practices the company is already leveraging.
I haven't attended Velocity before. It's pretty great so far.
Most striking: DevOps doesn't have empty buzzword cachet here. You can't just yell DevOps! -- or Docker! -- and get people excited. Bryan Liles, one of this morning's keynotes, set the tone with a hilarious slide I wish I had photographed: black background, white text, nothing more; and the white text just said, "Containers, containers, containers, containers, containers, containers, containers." He read it fast out loud; added a sentence or two about
systemd-nspawn; the crowd laughed; that was all. (Bryan's talk touched on another aspect of the 'people' side of DevOps, connecting two major issues I hadn't seen connected before. More on this later.) No Dockerwashing here.
More valuable: many talks (and vendors) take DevOps values and techniques for granted, then push hard to refocus on the user, no seriously, not just the idea of the user.
One result: the most serious and well-informed discussion of responsive images I've ever seen, without any mindless acceptance of the spec. Nobody here wants to hold up a (W3C/WHATWG) rule and say, "Here's how to follow it." Speakers and vendors are immediately pushing beyond the spec, refusing to rely exclusively on the picture element when something server-side, even something based on a really nice collection of user-agent data will do better.
Another result: Steve Souders' refreshingly honest critique of lazy web performance metrics -- particularly naive over-reliance on silly conflations of
onload with the page is actually useful, or even overenthusiastic response to the (legitimately awesome) navigation timing and resource timing APIs. What's wrong with identifying a DOM event with a performance index? Users don't care about DOM events. Asynchronous loading is lovely if you're only interested in does this resource block or not but can frustrate (or infuriate!) a user who sees all the uninteresting elements loading first. Congratulations, your CDN is so quick that your user gets to stare at your beautiful logo for five seconds while the 'reply to message' input box faaaaaaaaails to appear. Okay, you're asynchronously loading cached images. But your website is still slow. Or rather: your users can't do what they're waiting for.
Souders' talk was titled "Design + performance" in order, I think, to emphasize that user-facing stuff -- stuff designers handle, maybe pre-eminently -- is what really matters. Everyone knows that; but not everyone concludes that design has just as much of a stake in performance as dev and ops. (There's an interesting-looking book about this, published by O'Reilly of course. I haven't read it. Yesenia Perez-Cruz, who also spoke today, has done a lot of work on this problem from the design perspective, with a bit of extra emphasis on typography, I think. Well, time to start reading...)
Not that there isn't deep comms-and-browser-as-platform-level stuff going on here too. Two such talks I wish I had seen: Ilya Grigorik's presentation on the ServiceWorker API; and Ragnar Lonn & Daniel Steinberg[the curl guy! the http2 explained guy!]'s performance analysis of HTTP/2 vs. HTTP/1.1. No doubt the latter presentation was great, but no matter that we didn't catch it live: the results are now available in a public tool, which R&D announced today. (HTTP/2 sounds pretty awesome -- I dived into the topic via SE-Radio's excellent interview with Mark Nottingham, chair of the IETF HTTP Working Group, and I suggest you first listen to that podcast, then read this brief article, then dig into Daniel's deeper piece.)
Outside the sessions, I had a chance to chat with some super smart folks about some super interesting projects at a number of really innovative companies exhibiting here at the show. I'll be sharing more thoughts on their work, some more of the sessions, and a bit more on the concept of 'beyond DevOps' from both engineering and social perspectives. Stay tuned.
Opinions expressed by DZone contributors are their own.