RedHat Summit: Connect 2024
On the 9th of October we attended Red Hat summit for the day and attended their workshops on RHEL, Ansible and Openshift as well as other products, below we’ve summarised our key takeaways:
Pipelines, pipelines, pipelines – The keyword of the day was pipelines, a big part of the Red Hat suite is to be able to automate everything you do into workflows, “pipelines”, that allow your many actions to become one action. This revolved around event driven ansible which reads the events that VMs and systemd services output into message brokers, then actioning based on filters for those events. An example that was given was a webserver under load, once a load test was performed and the httpd service crashed, ansible actioned this by way of a “rulebook” that captured the failed service event, triggering the “playbook” to restart the service. It was taken further by additional logic counting how many events are triggered, so if the service crashed so many times a minute RedHat service interconnect launched an incident ticket that increased the resources on the VM to meet demand.
Could be that you pipe Monitoring tooling Alarms into Event Driven ansible and it can perform actions from those events. Like cleaning a C:\ drive, and then extending if space continues to grow!
In addition event driven ansible can drive automation to the point that instead of reporting events on windows or linux service crashes it could cause service stack restarts bringing the service back up and launching an incident ticket to look into it in hours
Ansible Automation Platform - The actions performed in the ansible workshops were facilitated by Ansible Automation Platform, we have the upstream version called AWX deployed into the environment to patch and baseline Linux VMs that take up our automated patching solution. AWX and Ansible Tower allow teams to manage playbooks, rulebooks, schedule jobs and monitoring for failed job metrics.
Ansible Lightspeed, AI for YAML – Have you ever thought, boy this YAML malarky sure is tedious, well do I have the product for you! Another exciting addition to the ansible space is lightspeed, a small scale LLM that allows you to train a language model to write YAML scripts for your Ansible playbooks and rulebooks in your own style. There are big applications for this to us, either from training the model based on existing customer deployments in order to deploy new customer YAML quicker or either allowing the predictive text to sense what block you’re writing and auto complete to save time.
Basically, allowing any normal person to create/edit task without needing to know all the nitty gritty YAML config. Which is the biggest hurdle operationally to get Ansible widely adopted.
vSphere is dead, long live the Red Hat - RedHat OpenShift is currently the flagship product they’re pushing for obvious reasons in the virtualisation space due to external factors with competitors. We attended a couple sessions on OpenShift, firstly one that went through OpenShift as a product and how we would go about moving over to it from VMWare environments. This session was the most informative on the product I feel, built for technical staff to have a look at the product CFOs might be pushing for because its cheaper. OpenShift was released in 2011 and is an orchestration and management layer to several open-source projects such as kubevirt, KVM and CRI-O.
VMs and Containers are both treated as ‘Pods’
This means Veeam Kasten would need to be deployed for backups! Also we only need to learn one set of cli/api calls to operate both containers and VMs making integrations easier.
RHEL licensing is absorbed into OpenShift licensing much like Hyper-V and Datacentre licensing, meaning any ‘Linux clusters’ would likely be cheaper running on OpenShift
OpenShift is the clear winner in a solution to run customer containers, the technology its built on is tried and tested for containers
AI is the answer, what is the question – In addition to Ansible Lightspeed Red Hat also pushed the Granite LLM, this is an interesting piece as it’s a small 30b model designed for smaller specific tasks. The premise is you deploy granite, train it to respond to specific inputs with information you’ve crawled though at certain intervals of the day (This is a generalisation of AI for what I think it may be useful for).
My thoughts, push granite to crawl and index our documentation and then have a front end where you prompt it like so “Get documentation for RDS Certificates”, it would take keywords like customer, certificates and RDS and give you the link for the documentation you’re looking for, if it cant find anything specific for the customer it’ll give you documentation without a customer tag, for example generic how tos that match the criteria for RDS and certificates.
Lastly, there are some mad people out there who completely redeploy their entire infrastructure on a monthly basis instead of patching! They patch the “gold” image and do testing, once testing is done they redeploy production, most impressive thing we heard all day!
Additional details and information can be found below:
https://www.redhat.com/en/summit/connect/emea/london-2024