Scaling Up Infrastructure Pipelines with Resource Management Abstractions

Today, RackN announced the release of Digital Rebar v4.8. With this release, Digital Rebar Resource Management Abstractions take center stage. 

You can read more about the release here. But let’s dive deeper into Resource Management Abstractions. 

Resource Management Abstractions

Infrastructure Pipelines are standardized, reusable processes for platform and infrastructure. Like CI/CD pipelines for code, they provide a consistent, predictable, and automated end-to-end process that provisions, updates, maintains, and decommissions infrastructure.  

They are composed of modular workflows that are orchestrated by a pipeline manager throughout their multi-cloud, on-premises or edge lifecycle. To do this requires the use of Resource Management Abstractions. The initial Digital Rebar abstractions were Machines and Workflows.

Machines

Machines are data center components such as servers, VMs, switches, etc. They are inventoried by the Digital Rebar Platform. Once the platform understands the capabilities of the Machine (ram, drivers, memory, disk space, etc), it is available to have a state applied to it.

Workflows

Workflows contain all the processes used for building end-to-end infrastructure pipelines. In the beginning, a machine must be configured to a specified state. Need another ESXi node for your cluster? Need a Windows server? A Rocky Linux server? Digital Rebar has standardized workflows for that.

This is how Digital Rebar has worked until now. And this is no small thing, we have customers running this in production at a 10,000+ Machine scale! Our  production-hardened automation library means you’re not searching  snippets of code  for foundational things like deploying servers.

But we needed more Resource Management Abstractions to make the jump to multi-cloud.

New Resource Management Abstractions in 4.8

We are introducing three new Resource Management Abstractions with this release: Resource Brokers,  Clusters, and Work Orders. 

Resource Brokers

A Resource Broker manages resource created by clusters. In other words, Resource Brokers provide a way to generically interface with other types of systems (bare metal pools, containers, or clouds. ). 

Clusters

Clusters are the highest level resource management abstraction level in Digital Rebar. A Cluster manages a collection of machines. To do that, it first requests machines from its assigned Resource Broker. Next the Cluster automatically coordinates an Infrastructure Pipeline to install platform software.  Once the install completes, the Cluster switches to Work Order mode to manage jobs and tasks against the running cluster.

If you need to grow or shrink the cluster, simply adjust the size of the Cluster, and Infrastructure Pipelines will be kicked off  to build or destroy machines. Each cluster has its own lifecycle that can be extended into other automation platforms like Kubernetes, vSphere, or SLURM.

Work Orders

Work orders take systems from a life-cycle state to a jobs execution stage, scheduling additional tasks along the lifecycle of the system without destroying or changing it.

Putting it all together with Resource Management Abstraction

  • Clusters are created by selecting a size, a target Resource Broker and an install pipeline (e.g.: Kubernetes).
  • Next, the Cluster Install Pipeline makes work order requests to the Resource Broker with target pipelines for each machine.  While the Resource Broker processes the work order, the Cluster can monitor the status of the machines in the cluster and perform other management tasks.
  • Meanwhile, the Resource Broker requests machines from a cloud (via templated Terraform plans) or from a bare metal server pool.  
  • As machines come online, each runs its own version of the requested pipeline.  
  • Since the cluster pipeline is monitoring the process, it can select machines for specific tasks or roles.
  • The lifecycle management continues after the cluster install process completes by shifting the cluster into work order mode.  
  • The cluster is able to take on asynchronous requests such as adding user accounts, running helm charts, or checking conformance.  
  • To resize or destroy the cluster, operators shift back into workflow mode and start the appropriate pipeline.

Demo Time!

This video shows the Cluster, Resource Broker, and Work Order features.

Resource Management Abstractions unlock the power of automation

These new architectural features enable some pretty cool things. What if I have some infrastructure for an app running in AWS that I want to try in Azure? Just build a new cluster with the Azure Resource Broker. Can I use the same Terraform scripts? Yes – and Digital Rebar can also help you see where the requirements may be different on the new cloud platform (e.g. case). 

This release makes it possible to write automation that autonomously builds the required architecture and assembles the necessary resources for platforms like Kubernetes, vCenter, SLURM in a consistent, multi-cloud way. But that post is for another day.

Real Talk

These new Digital Rebar Resource Management Abstractions will help you automate infrastructure during all parts of its lifecycle. 

Try it yourself:  https://rackn.com/trial/

Learn more about the Digital Rebar v4.7 Launch:

Spread the word. Share this post!

1 Comments

  1. Pingback: Pump up your Infrastructure Pipelines with an injection of Ansible - RackN

Leave a Reply