The traditional IT systems administrator role is an endangered species due to the emergence of cloud computing...
and new forms of application development. That's why it's incumbent on those in the field to find new ways to remain relevant in their companies.
That hard reality was broached by Google's Dave Rensin during a panel discussion on innovation in the digital age at a Rackspace roadshow in New York. Rensin, who heads Google's customer reliability engineering team, should know a thing or two about this subject: He helps large enterprises shift to Google Cloud Platform and more modern IT practices.
SearchCloudComputing spoke with Rensin after the panel discussion about how the systems administrator role can remain relevant in the workplace. Later at the event, we asked the same question to a trio of executives also on hand for the conference: John Engates, CTO of Rackspace; Brannon Lacey, general manager and vice president of applications and platforms at Rackspace; and Ian Andrews, vice president of products at Pivotal Software. These two companies are fully immersed in the changing IT landscape, both on premises and in the public cloud. Here's their advice.
Editor's note: Responses were edited for brevity and clarity.
Dave Rensin: I started as a sys admin with SunOS and VAX ... That world is leaving rapidly. If you're not in your 40s or living out the tail end of your career, you've got to bite the bullet and learn to write software.
You don't have to be a tremendous software engineer, but you've got to write enough software that if you encounter a problem that annoys you, you have the skills to automate it away. If you don't have that, you're in kind of a dead end. You're going to see the value of what you do diminish over time in the marketplace. It's not fun to say that to people; that's not a pleasant thing to say, but it's just the truth.
You don't have to go to college and get a CS [computer science] degree. There's no shortage of online things you can do to learn reasonable basic Python programming or Go programming. You don't have to go all the way to C++ or anything. Every time you do a thing, ask yourself: How do I automate this? How could I repeatably, provably automate it?
Sometimes, I'll meet admins who say not everything can be automated. That's a mindset change. Then, I say to them: 'Could you write a playbook for it? So, you've written a script. Why can't you make the computer follow the script?' That's really the essence of it.
It's a lot like 25 years ago. You'd meet a software developer that was a procedural programmer. And you tell them, 'Look, I get it, I'm sorry, but software is complex and you need to learn object-oriented.' [If] you don't, the window of your opportunity is going to continue to shrink to more niche and niche things.
John Engates: Every one of our top sys admins at Rackspace -- the ones that have risen to the top -- have crossed that chasm of actually going from being a sys admin to being somebody that can write a bit of code. It doesn't mean you have to go write the core application, but you have to be able to code enough or be able to communicate in code enough, and be able to automate code and be able to drive something like Cloud Foundry, instead of being a bystander.
Most of the good sys admins have picked up a bit of code, and it's beyond just the old world of shell script and a little bit of Perl. And when they sit side by side [with developers], it rubs off. They start to really absorb it. Don't be afraid of it, because most of these guys are sharp, and they can pick it up and they can learn, but the one that refuses to is going to be stuck in the old world.
That's the direction things are going. When you get a certification at any of the major cloud providers, there's a bit of code that's required. I hope that's what comes through when Dave [Resnin] says that sys admin's going away.
Ian Andrews: I totally agree with that, and I would add to it the phrase du jour, infrastructure as code. You used to have your sys admins sit in the cold corner of the data center and actually touch physical machines. They probably helped rack them and run cables, and they know each machine by heart. In the public cloud context, certainly, that role is entirely gone, unless you're employed by one of the big public clouds. If your career path is inside of a large enterprise, that skill set is going to be less and less needed over time.
You then start to think about provisioning infrastructure, not by physically racking it, but by running a Terraform or [Amazon Web Services] CloudFormation script that provisions infrastructure. And what's important [to do] that? I need to understand Git or some sort of source version-control system, because that's how I build the environment. That's how I update and make changes to it. So, there's a lot of tooling, even if you're not a Java dev, that becomes really important.
Ian Andrewsvice president of products, Pivotal Software
Continuous integration and delivery of platform infrastructure is the next wave, [which is] anything around automated testing, canary deployments, smoke tests. That whole area has been the exclusive domain of app developers. Blue-green deployment across your infrastructure is where the state-of-the-art teams inside of Google who were provisioning that infrastructure are working, and that ends up coming out to the enterprise.
The other thing that's a big mental shift here is that IT operations and sys admins never knew what was running on the servers. Even today, it's not unusual for Pivotal to hold a meeting where, for the first time, ops and developers are talking to each other. I think if you can do one important thing besides learn to code, it would be go find out what these applications are and what they are doing for your business.
If you're having that conversation, then you can be extremely valuable. You've got technical expertise, you've got knowledge and concepts the other people at the table don't have. The devs aren't going to be infrastructure experts. As much as people want to say DevOps all day long, you're still going to have specialization there, but you can't participate in the conversation if you can't understand the apps.
Brannon Lacey: Sys ads and network ads make really good analysts in a security operation. Security is becoming a growing tidal wave, and whether or not an MSSP [managed security service provider] is brought into an organization or not, the ability to understand what's happening in the environment and leveraging network [aligns well with] sys ad skill sets. I've got an SOC [system on a chip] filled with ex-sys ads and network admins; it's the best foundation.
The new blue-collar job is coding. I think everybody is going to be coding, so absolutely coding goes without saying, but if you summarize all of it: It's understanding the business implications of what you're doing from an IT perspective. Go learn the application and coding aspects of it. Go learn the security.
Help dev and ops teams find a common language
Three cloud computing skills to hone in 2017
Learn these important skills, tools for infrastructure as a code