Return to site

DevOps Needs System Models

Modeling for Operations

Operations teams need to model their systems.

I am astonished by how many IT Ops teams don’t really have a good diagram of their system at any level, let alone a real model. They literally have a very limited idea of what they have, or are running. As they say:

This is no way to run a railroad

No logical model nor diagrams of every layer (physical, network, data flow, interaction, etc.) means endless problems, guessing, and uncertainty. Plus it’s very hard to create good automation if the system’s fundamental architecture is not modeled and available in some digital format.

Infrastructure-As-Code makes this worse as the ‘model’ is buried in a zillion config files, somewhere in git, strewn across config files, playbooks, Cloud Formation, and AMIs.

What a mess ...

All this is much, much worse as we move to MicroServices where there are a lot more moving parts, often with murky and rapidly changing connections, with pieces coming and going such that no one really knows what's where, there, anywhere.

How people manage this, I’ll never know - there are so many challenges you may not even be aware of, until you've used a model-driven management system and seen what it can do for you.

What do to ?

Build a model. Of the whole system. Top-to-bottom. Of EVERY component. Including the configuration of every component. And keep it updated all the time.

This is very difficult to do. So of course we did it for you. 

OpsStack contains a full-stack detailed model of every part of your system, from the cloud to the OS to each service, micro-service, and more. Including the full configuration of each, from AWS options to Linux users to cron jobs to MySQL buffer size to SSL TLS ciphers and more. We know everything about everything.

How do we do this ?

We started with TOSCA, the Topology & Orchestration Specification for Cloud Applications, from the OASIS standards body. TOSCA is a similar to AWS Cloud Formation, but far more powerful and extensible, partly because it also includes orchestration, relationships, and much more.

Starting with TOSCA, we heavily extended and ‘improved’ it to cover the full-stack and entire breadth of cloud operations, including deep configuration and practical components to enable forward, reverse and change engineering of cloud systems.

For Forward Engineering, we have Design Editors to build up a system from scratch or templates so you can design your cloud or physical server infrastructure, then add in OS elements, then services such as Nginx, PHP, MySQL, Redis, and more - all configuration is done in a single place. Everything is rendered into the TOSCA model, including all dependencies, relationships, and connections.

From this model, you can build a real system, as OpsStack can create the AWS components and then automatically install all the required services, configure them, setup monitoring, and more. This is the world’s only full-stack design, model, and build system.

For Reverse Engineering, we talk to a cloud like AWS and your servers using OpsStack CMDB Collectors, and EXTRACT your current architecture and full-stack configuration, including all AWS options, services, users, vhosts, databases, and config items. All this goes into the TOSCA model.

From the Forward or Reverse Engineered TOSCA model, you can make changes and SAFELY push these back to the existing system. This lets you change any attribute of anything and push it back into your running system. Or clone to a new system. This is the world’s only full-stack discover, reverse-engineer, model, and change system.

Safely change your system

With these models you can now visualize, monitor, and understand your system. Then we add automation, dependancy-based alerting, and auto-scaling plus features to manage, check compliance, drive security, and much more. You can know where everything is, what it’s doing, and where problems are, all the time.

We even add in service-level monitoring, clustering, and real actionable alerting. On top of that, we add tuning and troubleshooting tools including AI-driven rule-engines, cluster analysis, and more, all driven from the model.

The model is your friend

Get a model today.

Get more info on OpsStack at www.OpsStack.io