In the first part of this series, I explored the factors that have lead DevOps to where it is today. From a three to six month deployment period to regular weekly (or even daily) deployments, DevOps has certainly changed the IT landscape as we know it. Now, with the growing adoption of a DevOps mindset, the IT world is seeing a huge growth in software companies created specifically for DevOps professionals.
The explosion of monitoring tools and evolution of the software engineer role
With the added complexity of SaaS software and the rapid innovation of the platform, issues occur more frequently. With the adoption of unified cloud infrastructures more and more of these issues are application layer in nature. Service providers like Amazon are now consuming the need for foundational monitoring from tools such as Nagios at the IaaS (Infrastructure-as-a-Service) layer. Today, problems are popping up at a higher level in the stack and, monitored by tools such as NewRelic, that output is directly destined for software engineers. The ability to know if something is working correctly is now a comparative operation performed by the engineers who wrote the software.
The changing landscape of the NOC
If getting software built and into production was the nirvana that everyone had been marching toward, the NOC (Network Operations Center) was the manual switchboard of the past standing right in the way.
Traditionally, the NOC has been the catchall command center for people who monitor the systems that monitor the software. NOCs were designed to be the first line of defense against technical problems, customer problems, and general communication of issues to the wider company. Historically, NOCs have had an infrastructure bias as the software higher in the stack changed less often. But as the true infrastructure layer of SaaS companies began to move away from the company and into IaaS platforms, that need was receding. At the same time, as application deployments were increasing in frequency, often first responders that were not highly technical were adding drag in system resolution times.
Today, companies are looking for ways to direct problems straight to the “makers” that created that component of software in the first place. With the speed of SaaS businesses, the company can no longer afford for the first responder to be anyone other than the person or group that created the logic in the first place.
If the NOC of the past had two critical functions, solving technical problems fast and communicating with internal and external stakeholders, suddenly half of their charter was starting to fail dramatically. This led to the explosion of software like VictorOps and monitoring tools like New Relic and AppDynamics to help diagnose these problems quickly.
Operations is dead, long live DevOps
In all this time a lot of things have changed for operations people. While a technical operations manager in the past might hire system admins to do all the manual configurations, now they are looking for team members with coding skills. While understanding hardware was a requirement in the past, now suddenly virtual servers that you can rent from Amazon, Rackspace or any of 30 other IaaS providers can be created (spun up) in an instant with no knowledge of hardware at all. In the past a system admin had little to no knowledge of how the application software worked, but now that is their primary job description. In short, today you want to hire a software developer of a different skill set – one that codes the infrastructure and understands the application. While the NOC was once your first line of defense against system problems, now the team that wrote the software is the only defense necessary for technical matters.
For many, the end game has been realized
Now, for more progressive companies, the big gain has been realized. After literally a decade of innovation, for the first time it is easier to fix a production problem in-stage, rebuild and redeploy (all automatically) than it would be to fix the problem in production. Today the reality of being able to execute an Agile development sprint front to back is possible.
