LARA [Labyrinth Labs Reference Architecture] keeps evolving, and this time, it sings. Literally.
Version 8.1 of our cloud-native platform is out, and while it started as a simple bugfix patch [8.0.2], it quickly became something more - a feature-packed, stability-boosting, future-proofing release. And yes, we named it Smooth Operator [thanks, Carlos Sainz Jr. and the F1].
Let’s take a look at what’s new, what we fixed, and what it means for your infrastructure.
Quick Recap: How Did We Get Here?
- 640 hours of effort
- 27 issues closed
- 40 pull requests merged
- And one legendary moment when a seemingly innocent release turned into a full-blown feature drop.
We’re now shipping new LARA versions roughly every 4 months, which is faster than ever. Think of it like a Kubernetes-powered Pendolino train - just hopefully without fallen trees on the track.
What’s new in LARA 8.1?
Kubernetes 1.32 - aligned with AWS
We’ve caught up with AWS and bumped LARA to support EKS 1.32. That means standard support until March next year - giving platform teams some well-deserved vacation time [just kidding] without scrambling for emergency upgrades.
This version also introduces:
- nftables support - paving the way for improved networking performance and stability
- Optimized cross-AZ traffic routing - lowering costs and boosting efficiency
- Future-readiness for Amazon Linux 2023 - plus better support for registry caching [useful when internet egress is limited]
Karpenter upgraded to 1.4
Karpenter - our go-to Kubernetes autoscaler - has been migrated from beta to a stable 1.4 release. No massive changes, but one more piece of infra that’s future-proof and cleaner to maintain.
AWS Aurora Postgres upgrades
We bumped default Postgres versions to stay clear of AWS Extended Support [which, as we know, means extra charges]. The new default is supported until February 2028. Less worry, fewer upgrade headaches down the road.
We also added support for AWS ElastiCache Valkey - a high-performance open-source Redis alternative that offers more power at lower cost. Great fit for caching-heavy use cases.
kube-scheduler as a native addon
Our friends at Pixel Federation needed kube-scheduler support as part of their move to LARA. We made it a native addon.
Cleanup Time: Killing Off Old Skeletons
Goodbye, Helm bug nightmare
We finally resolved a long-standing Helm-related issue in ArgoCD that had been haunting our deployments. The fix unblocked our path to ArgoCD 3 in the future and made a lot of people very happy - especially those managing Helm charts with custom app packages.
Farewell, Bitnami
Bitnami’s sudden shift towards paid image registries caught many off guard - including us. They moved key images out of reach, and gave the community just two months to adapt.
We are currently in the transition period and we believe that in LARA 9 we will be completely Bitnami-free.
Bonus Upgrades in 8.1
- Kafka [AWS MSK] upgraded from 2.8.1 → 3.6 to stay ahead of forced deprecations
- Added metrics collection for Kafka → Prometheus
- Expanded HA support for ArgoCD workloads - ensuring the UI stays responsive even during pod rescheduling
Looking ahead: LARA 9
We’re already working on the next release. Early signs? It’ll be one of our most feature-packed updates ever. Here’s a taste:
- EKS 1.33
- Major internal refactoring powered by a Terraform version upgrade
[because LARA’s complexity started making old Terraform sweat] - Upcoming support for OpenTelemetry
- Atlantis
- Prometheus 3
- Plans for a more developer-friendly lifecycle
- OpenTofu compatibility is in the pipeline - just in case
Conclusion
LARA 8.1 is another step forward - smoother, faster, smarter.
We fixed long-standing pains, laid groundwork for the future, and made sure your infra won’t fall apart when someone at Broadcom has a bad Monday.
And if you’re not using Labyrinth Labs LARA yet? You should be.
We’re building one of the most advanced, open, and production-ready cloud platforms in the world - and we’ve got the customers [and the scars] to prove it.
