Just what everyone needs right now – an HPE chat bot wrapped in a Docker container, right?
IT titan talks up ChatOps, because email sucks
Hewlett Packard Enterprise has a bot for you. The company has taken GitHub's Hubot code and wrapped it in a Docker container, to promote the conceit of ChatOps, a model for combining communication tools with DevOps automation.
At the DevOps Enterprise Summit in San Francisco on Monday, Daniel Perez, software engineer at HPE, provided a demonstration of HPE's Hubot, named Hammer.
Hammer can handle data lookups, create graphs, run automations, provide application metrics, carry out commands, and even tell jokes, though none were offered during the demonstration.
Written in Node.js and CoffeeScript, Hammer stores data using Redis Brain and Mongo, and it is integrated with GitHub Enterprise, Electric Flow, and Flowdock. It aspires to be an automated assistant, like Cortana or Siri, but for interacting with IT services.
"You want to automate things that make sense, but you really shouldn't be automating everything," said Perez.
DevOps is not simply the automation of application development, deployment, and management. It's a process for continuous operational improvement that involves automation.
If you use the term "ChatOps," then add messaging applications to the mix.
In an interview with The Register following the bot presentation, Ashish Kuthalia, senior director of marketing and strategy for Hewlett-Packard Enterprise DevOps, said he expects ChatOps will become even more compelling as natural language processing makes bots smarter.
"The advantage in IT of a system like ChatOps is there are certain things you do over and over and over again," he said. "And we spend a lot of resources and money and time solving the same problem several times a year or month. And that's where these systems start to get a little bit intelligent. They recognize your patterns."
John Jeremiah, technology evangelist at TechBeacon, an HPE publication focused on enterprise IT, joined Perez for the presentation and spoke with The Register afterwards.
"In the demonstration where Hammer was interfacing with itself, that's nice," said Jeremiah. "But Hammer is going to get smarter. You have Siri, right? Or Google. Or whatever. But if you look at the power of those systems, the ability to do contextual interaction with people, that's where the bots are going."
Jeremiah argues that email is broken and chat applications have the potential to become the intermediaries between people, machines, and software scripts.
"We all hate email," Jeremiah insisted. "I haven't met a single person who loves email. We hate email because it's serial, it's disconnected [from other applications], it's overwhelming, and it's impossible to manage.
"What ChatOps gives us ... rather than having a long, threaded email discussion, [is to have] have one team, one discussion, one place for context. So we bring the bot to the room. And if the bot is able to participate in the context of what's going on, I don't have to ask the bot to go get the metrics for how the server is performing. A bot will be part of the conversation and will offer that up because it's smart enough to understand the context."
Bots can be expected to get smarter, though they're still limited by natural language parsing. At least the range of commands likely to be directed at Hammer and its Hubot kin won't be so obscure as to baffle parsing algorithms.
"We strongly believe that ChatOps really is a key enabler of DevOps through the use of persistent chat," said Perez.
"Being able to reference historical data in the chat room, being able to bring key developers into conversations and execute on-demand live automations ... makes it very powerful. It allows you to develop at a much quicker pace, and provides accountability for the team, knowing who ran what command at what time." ®