Working on Triton in the lab, what’s on the horizon

January 27, 2016

As we’ve talked about before, a few of us in Dell’s CTO group have recently been working with our friends at Joyent.   This effort is a part of the consideration of platforms capable of intelligently deploying workloads to all major infrastructure flavors – bare-metal, virtual machine, and container.

Today’s post on this topic comes to us complements of Glen Campbell — no, not that one, this one:

Glen has recently come from the field to join our merry band in the Office of the CTO.  He will be a part of the Open Source Cloud team looking at viable upstream OSS technologies across infrastructure, OS, applications, and operations.

Here is what Glen had to say:

What’s a Triton?

Joyent’s Triton Elastic Container Infrastructure, a Private Cloud variant of the Joyent Elastic Container Service PublicTriton slide

Cloud, allows customers to take advantage of the technologies and scale Joyent leverages in their Public Cloud.

On the Triton Elastic Container Infrastructure (which I’ll call “Triton” from now on) bare-metal workloads are intelligently sequestered via the use of the “Zones” capabilities of SmartOS.   Virtual machines are deployed via the leveraged KVM hypervisor in SmartOS, and Docker containers are deployed via the Docker Remote API Implementation for Triton and the use of the Docker or Docker Compose CLIs.

What’s the Dell/Joyent team doing?

As part of interacting with Triton we are working to deploy a Dell application, our Active System Manager (ASM), as a series of connected containers.

The work with Triton will encompass both Administrative and Operative efforts:

Administrative

  • Investigate user password-based authentication via LDAP/Active Directory
    • in conjunction with SSH key-based authentication for CLI work

Operative

  • Use of:
    • Admin web UI and User Portal to deploy single/multi-tier applications
    • Joyent Smart Data Center (SDC) node.js client to deploy from remote CLI
      • Newer Triton node client to see next-gen of “sdc-X” tools
  • Docker Compose
    • build a multi-tier Docker application via Docker Compose, deploy on Triton via its Docker Remote API endpoint
  • Triton Trident…
    • deploy a 3-tier application composed of:
      • Zone-controlled bare-metal tier (db – MySQL)
      • Docker-controlled container tier (app – Tomcat)
      • VM-based tier (presentation – nginx)
    • Dell Active System Manager — a work in progress
      • aligning with Dell’s internal development and product group to establish a container architecture for the application

Stay tuned

Our test environment has been created and the Triton platform has been deployed.  Follow-on blog posts will cover basic architecture of the environment and the work to accomplish the Admin and Ops tasks above.  Stay tuned!

Extra-credit reading

Pau for now…


Intro: Setting up Joyent’s Triton in Dell’s CTO lab

January 18, 2016

A while back I tweeted how we had begun setting up a mini-instance of Joyent’s Triton in our Dell CTO lab.  Triton is Joyent’s elastic container infrastructure that runs on their cloud, a private cloud or both.  This cloud platform includes OS and machine virtualization (e.g. Docker with regards to the former and typical VMs under KVM for the latter).

About a week ago we got the platform set up about and I grabbed sometime with Don Walker of Dell’s enterprise CTO office to tell us about it.

In this first of three videos, Don gives an overview of the work Dell is doing with Joyent.  He describes what we’ve set up in the lab and talks about where we hope to take it.

Some of the ground Don covers

  • Don’s focus on Open Source Cloud eg Open Stack, containers, cloud networking and storage solutions
  • What the enterprise CTO office does
  • What we’re doing with Joyent: evaluating Triton and the process of taking existing products and put them into microservices and containers.
  • Looking at Dell’s ASM (Active System Manager) and what it means to refactor for microservices and containers
  • Overview of what was set up in the lab: a minimalist 2 node instance consisting of head and compute nodes.

Extra credit reading

Pau for now…


%d bloggers like this: