ArchiTECHt Daily: Netflix shines a light on 'ChatOps' (and open sources its Slackbot)

If you’re enjoying this newsletter, please feel free to share it via social media or good, old-fashio
ARCHITECHT
ArchiTECHt Daily: Netflix shines a light on 'ChatOps' (and open sources its Slackbot)
By ARCHITECHT • Issue #13
If you’re enjoying this newsletter, please feel free to share it via social media or good, old-fashioned word of mouth.
On Tuesday, Netflix open sourced a chatbot—technically a Slackbot—for managing the company’s expansive GitHub environment. Called HubCommander, the bot lets developers automate their interactions with GitHub via Slack command, while also letting the company enforce and standardize its permissions protocol.
It’s interesting enough alone, but as I wrote on ArchiTECHt yesterday, Netflix is not the first company to adopt what some are referring to as “ChatOps.” In fact, the term goes back at least a couple years, with the most famous user being GitHub itself, via its popular Hubot system. To hear folks from GitHub tell it, Hubot lets engineers do everything from chat (naturally) to migrate workloads onto new infrastructure.
“Basically, everything you could ever possibly imagine to do in our infrastructure, you can do via Hubot,” GitHub’s Sam Lambert told me in 2015.
I have no idea the overall adoption rate of ChatOps today, but I’ll assume it’s limited to a relatively small number of tech-savvy companies—the Netflix- and GitHub-types, as well as startups modeled in their images. However, like so many other things that might initially appear foreign (DevOps and microservices, for example), ChatOps seems poised to make its way to the mainstream.
There are lots of reasons for why that is, but two big ones are its natural symbiosis with the type programmable infrastructure that began with AWS and carries on with containers, and the fact that chat logs are actually a relatively simple way to figure out who did what and when. A third might be the seemingly unstoppable force that is Slack, and all the competition it inspires along the way.
If I were Slack, Atlassian, Microsoft or any other number of companies with a horse in this race, I would start trying to own ChatOps now, while it’s still flying under the radar for most folks. Own the term, harden the technology, and get ready to make IT departments’ collective heads explode yet again.

HubCommander. Source: Netflix
HubCommander. Source: Netflix
Around the web: Artificial intelligence
Long story, short: TensorFlow models can now compute over different sizes and types of data without wasting resources and bogging down.
The idea of analyzing calls is fine in theory (in fact, I’m fairly certain Chorus.ai isn’t the first to do it), but a lot of people don’t like being recorded and some laws prohibit it without permission. 
The idea of custom vocabularies and commands in voice-controlled apps is kind of awesome, especially for gaming. I could see ChatOps going VoiceOps at some point, too.
More deep learning, of course. This has some definite utility for poorly show cellphone pics, but also security footage. Legal implications of “artificial” images notwithstanding …
The rich at Goldman get richer, the less-rich look for new employers as software handle the heavy lifting.
California lawmakers will now be recorded, with machine learning systems creating transcripts of what each one says on the assembly floor.
Around the web: Cloud and infrastructure
It doesn’t have to be just about the revenue, though. If Snap is innovative enough, Google could actually learn a thing or two about new services others customers might want, or other ways to tweak its platform. I would imagine Netflix was a boon for AWS aside from all the money it paid.
fortune.com  •  Share
That is a tough market to crack, especially with all the big buyers designing their own stuff and Intel seemingly embracing new data center workloads. This could be fun to watch.
www.wsj.com  •  Share
If Facebook is a customer, it must be doing something right.
Bias aside (I used to work a Mesosphere), this is arguably a more strategic deal for HPE, which needs a good story on container orchestration and cloud computing.
Source: Facebook
Source: Facebook
Around the web: All things data
Another Hive workload bites the dust.
Good on the players’ union for getting this into the collective bargaining agreement. I suspect consumers won’t be so lucky when insurance companies decide they want to do this.
This is an interesting explanation of how Slack built its search engine—unique because of each user’s narrow universe of available content—as well as another Spark use case.
I can’t imagine this moves the needle much for Docker adoption, but it’s a good selling point for MapR as customers move to build data-driven apps using containers.
Hopefully, open data sets on personal attacks can help other platforms get better at identifying and dealing with them in real time.
Can a new technology from a relatively unknown company dislodge better known real-time engines with bigger communities? Even if it’s better, the big fear is where it will be 3 years from now.
Did you enjoy this issue?
ARCHITECHT
The most interesting news, analysis, blog posts and research in cloud computing, artificial intelligence and software engineering. Delivered daily to your inbox. Curated by Derrick Harris. Check out the Architecht site at https://architecht.io
Carefully curated by ARCHITECHT with Revue. If you were forwarded this newsletter and you like it, you can subscribe here. If you don't want these updates anymore, please unsubscribe here.