Doing one thing at a time is so…nineties. We now have to do everything, all the time, like the guy texting as he walked past me into a ladies’ room in O’Hare Airport.
But that’s a different story.
The big emerging trend in IT multitasking is DevOps. It means combining what used to be the sequential tasks of creating an application (development) and keeping it running (operations.) Doing both in parallel should allow businesses to roll out new applications and services more quickly. That’s essential when a photo sharing site like Instagram needs to add 1 million users in 12 hours, users expect constantly updated mobile applications, and popular Web sites do continuous “A/B” testing to see if users like the scroll bar looking like this or like that.
Beyond Process Change
You might think DevOps is largely a “soft skills” story – how to get often warring development and operations teams to play nice. Development, after all, is paid to get cool new apps out the door quickly. Operations is paid to slow down and make sure they work right and are secure. And there are, indeed, plenty of good stories for marketers to tell about consultants who can do the necessary training and process change.
But it turns out there’s also a big technology story. Operations, after all, collects reams of log files and other data that track the operation of everything from Web servers to load balancers. By feeding that data back to the developers, in real time, they can tweak their applications and system architectures to avoid slowdowns, and adapt user interfaces based on what’s hot from the Web analytics that week.
This has raised the profiles of vendors such as Splunk, whose software monitors and analyses “everything from customer clickstreams and transactions to network activity to call records.” This can be used, among other purposes, “to debug and troubleshoot applications during development and test cycles.” Likewise, the CA LISA software suite from CA Technologies (one of my clients) simulates production environments to help multiple development teams work in parallel and manage test environments, another important part of the DevOps process,
So we know there’s a tech story to tell here. But in my conversation with vendors I’m finding some common challenges:
- Many customers either don’t know what DevOps is, or think it is hype. Define DevOps carefully and put it in context of related buzzwords like agile and open source. How you position all these trends isn’t as important as being clear in how they relate. Case studies of how DevOps has scaled securely in the real world will also help win over skeptics.
- Especially if you provide products or services on the data analysis side, make sure you explain exactly how you fit into DevOps process. This is a classic opportunity to define the conversation around an emerging market space by being first to explain it. (One sign of confusion: One of my clients described Splunk as a leader in the DevOps space, but Splunk itself doesn’t seem to agree, as a search for “DevOps” on their site yielded no hits.)
- Again, if you play in the data warehouse/data analysis/query tools space of DevOps, make sure you explain you’re analyzing machine data from the IT infrastructure, and not business data like when to put the beer next to the diapers to sell more of each. (A classic “Big Data” insight which, by the way, may have been ignored.)
So is DevOps real? Everyone from rocket scientist billionaires at social Web sites to somewhat staider outfits like the German Post Office say yes. Others will take more convincing. Ladies and gents, start your explanation engines.
Like this post? Subscribe to my RSS feed and get loads more!