A Baltimore transplant to Austin, Rob thinks about ways of building scale infrastructure for the clouds using Agile processes. He sat on the OpenStack Foundation board for four years. He co-founded RackN enable software that creates hyperscale converged infrastructure.
IPUs, intelligent processing units, are also known as smart NICs, side cars, or supervisory computers. A well known example is Amazon’s Nitro. We discuss the impacts of these supervisory processors, and how they can change the industry.
This is clearly the trend of the future. Building supervisory systems as additional processing capability into our core servers that then abstract out how the bus is interfaced how the routine peripherals, network activity, GPU or storage is abstracted in these systems.
BUT will they create a new way to manage heterogeneity and diversity inside of our hardware and server ecosystems? Can they have a dramatic impact on edge computing?
The “right to repair” is a really thorny and political issue! We talked a lot about John Deere, Apple and Tesla not letting people fix the products they’ve bought from those companies. We have a lot of questions! Why they do that? What the challenges are with RTR? How we could avoid them? What pressures keep us coming back to companies that are offering goods that we don’t have the ability to repair?
How can cloud economics of hyperscalar clouds be used to ended or limit their control? We discussed the state of the cloud ecosystem with a focus on where its going to go. After a full, rich, and dynamic conversation, we came back to security security, software (as opposed to SaaS), owning your own infrastructure, and the ROI of that infrastructure.
To understand CI/CD of infrastructure, you have to actually understand how the systems work together. In this podcast, we talked about low code, no code and how you would do integration testing for that. We also explore the idea of unit tests and code coverage for CI/CD presents challenges.
How do we secure service meshes for zero trust? Especially since we don’t know how they’re going to be used or integrated?
We went very deep into what it takes to secure service mesh. That included what type of application frameworks are going to be required to provide multi-service infrastructures that are secure and trusted in a performing way.
Can you composing infrastructure tasks: provisioning, configuration, security and monitoring, all together? Basically a Linux Pipe for automation? Maybe!
We talked about how hard it is to build robust, resilient infrastructure. The theme was supposed to be what could go wrong and finding the weak links in our infrastructure automation. Fundamentally we think that it’s all weak links!
We had a good conversation about how we actually build and automate robust applications. But we didn’t come to consensus.
How can GitOps and Infrastructure as Code scale your team’s ability to write automation? We answer how to take your DevOps to the next level. We really dig into what the automation challenges are and how you can do better as you build an integrated infrastructure system.
And of course, we did talk about Env0 and how Env0 solves the problem. And you will learn a lot about Env0 during this podcast.
Tim and I did discuss what it takes to build real infrastructure, automation pipelines,
Our Service Mesh discussion leaned heavily into the needs around edge infrastructure because there are so many missing parts for the edge deployment systems,
When we started talking about service meshes, we really realize is that the actual control plane, communications grid and security for edge are not defined enough for us to layer on what has become sort of a standard in cloud deployments of service mesh into that discussion.
How we got there, how we discussed it, and the components of why that’s important, is much more interesting than the conclusion itself.