Why the Cloud is Driving IT Re-Tooling
Dec 04, 2019
The provisioning and movement of infrastructure in the cloud is so dynamic that the IT tools and processes of 15 years ago are no longer applicable. Most large enterprises now find themselves in a transformative phase of IT, where it is vital to define infrastructure, security, policy, and network automation as code.
- Dave McJannetCEO, HashiCorp
How we provision compute capacity is different in the cloud world versus the traditional worlds that we are accustomed to.
In my current environment, I probably have a dedicated fleet of servers. I may have hundreds of thousands of servers, and we all have our own. That's my server, that's your server. We may pool them to a degree using tools like virtualization where I create a common layer atop a set of servers that look the same.
But when I need compute capacity, for example, if I'm at a development team that wants to deliver an application, what I have to do is file a ticket for someone to stand up a virtual machine on that computer, on that server. Once that is completed the ticket's closed, and I can start working on it.
That is appropriate for the kinds of applications that people had me building for the last 10 years, but the new applications that people are building by and large are going to be much more spiky in nature in terms of their consumption patterns.
The application I was building before, which may have been designed for hundreds or a few thousand people on vSphere is now an application that could look like Fortnite or Pokémon Go. Or maybe it's a big analytics application that needs to be spun up.
In this newer world, what I need to be able to do is spin up compute on-demand. I need to be able to do it in a self-service manner because I need to be able to take advantage of the fact that I may have 100,000 people using this application at noon on a Wednesday and 1,000 using it at midnight.
What cloud allows you to do is be much more nimble in the use of infrastructure. So rather than dedicated, owned infrastructure, you're now essentially renting infrastructure. Rather than filing a ticket for someone to open up a virtual machine on top of my own infrastructure, I'm now spinning up compute capacity on-demand.
The implication of that is the tooling and the workflow that I've used for the last 15 years to solve that problem is no longer applicable to this new problem. I think that's one of the big transitions we're seeing is people adopting the notion of codification of infrastructure so I can spin it up in a repeatable way rather than asking my virtual machine administrator to click 500 times to spin up the 500 servers.