Introducing Upgrade Orchestration: Safe Fleet Operations for Enterprise Platforms

Cat Morris
1 day ago
6 min read

There comes a point in every platform team's journey when dealing with their success becomes their next challenge. You've used Syntasso Kratix Enterprise (SKE) to standardise how developers consume platform capabilities. You've reduced manual work. Hundreds, perhaps thousands, of resources are now managed consistently through Promises. And then it's time to upgrade them.

A new PostgreSQL version is available. A security patch needs to be rolled out. An internal platform capability has evolved. The question is no longer can you upgrade your platform. It's how.

Upgrade everything at once, and a mistake can ripple across your organisation. Upgrade resources individually, and your platform team becomes the bottleneck. Delay upgrades altogether, and technical debt and security vulnerabilities quietly accumulate.

Enterprise platforms shouldn't force you to choose between speed and safety.

Today, we're introducing Upgrade Orchestration in the Syntasso Kratix Enterprise Platform Manager: a new capability that gives platform teams precise, declarative control over how platform changes are rolled out across an entire fleet.

Fleet management is only half the story

One of Kratix's biggest strengths has always been fleet management. Update a Promise, and every resource built from that Promise can evolve automatically. Instead of manually updating hundreds of individual services, the platform manages them as a single fleet.

That changes the economics of platform operations, but production environments introduce new requirements. Not every application has the same risk profile.

Customer-facing payment systems shouldn't necessarily upgrade alongside internal tooling. Production systems may only be allowed to change during approved maintenance windows. Security patches might need to be rolled out immediately to development environments, but only gradually in production.

As platforms mature, the need to govern changes across the fleet becomes increasingly important.

Declarative operations for platform upgrades

Upgrade Orchestration builds on Kratix's Promise Revisions and Resource Bindings to enable platform teams to control when and how changes are rolled out across the entire fleet.

Here's how it works:

You define an UpgradePlan that specifies which resources should be upgraded, in what order, and under what conditions.
- You decide if upgrades should roll out in batches, respect maintenance windows, or follow a schedule.
- You choose how many resources can upgrade in parallel.
When you are ready to execute that plan, you trigger an UpgradeJob, either manually or by providing rules in the UpgradePlan.

And the SKE Platform Manager handles the rest, respecting every constraint you've defined, and showing you exactly which resources are being upgraded in real time.

Creating a plan doesn't execute anything, so you can perfect your rules before making any changes. You can preview the run before it begins, allowing you to see exactly what will be affected before initiating the upgrade. And you can run the same plan over and over, whether that's a one-off change or a scheduled, automated rollout.

A demo of Upgrade Orchestration in action.

A real-world example of the SKE Upgrade Orchestrator

Imagine you've updated your PostgreSQL Promise to use the latest minor release version of the database. Four hundred database instances now have a newer revision available.

Without Upgrade Orchestration, your options are limited. Upgrade everything immediately and accept the risk, or manually coordinate upgrades across dozens of teams over several weeks.

With Upgrade Orchestration, the process becomes predictable:

Development databases are upgraded overnight.
Internal applications are upgraded first.
Customer-facing services roll out in carefully controlled batches during weekend maintenance windows.
If the canary batch encounters a problem, the rollout stops automatically before the wider fleet is affected.

The same upgrade plan can then be reused every time PostgreSQL evolves.

Built for enterprise platform operations: Key features

The SKE Platform Manager Upgrade Orchestrator includes:

Canary deployments, built in: Define your resources in batches. The first batch acts as a canary and is upgraded first. Only when those resources have been successfully upgraded do you move to the next batch.
- This means if something goes wrong, you have already caught it before the blast radius widens.
- You catch problems early, before they affect your entire fleet.
Maintenance and change freeze windows: Specify time windows when upgrades are and aren't allowed.
- Only upgrade resources in production between midnight and 4am on weekends, or apply change-freeze windows that mean nothing can change on Black Friday.
- The UpgradePlan respects those constraints and re-queues upgrades that fall outside the window.
- This means upgrades happen when you and your business decide. You have added guardrails against breaking production when customers would be impacted.
Full traceability and control: Before executing a plan, you can see exactly which resources will be affected.
- The upgrade list is locked in at execution time, so you know you won't hit new resources created mid-upgrade.
- Every upgrade run is recorded. You can see when it started, which resources succeeded, which ones failed, and who triggered it.
- This means you can review the impact before you press go and have a complete audit trail. You know what is going to change, and then if something breaks, you know exactly what changed and when.
Scheduled upgrades, hands-off if you want them: Define a schedule in your plan, and the orchestration executes automatically.
- You can run security patches every Tuesday at 2 am, or you can upgrade development environments nightly. The plan can handle recurring upgrades without you having to manually kick off each run.
- Now, routine upgrades happen without a second thought. Your team stays focused on platform features, not upgrade mechanics.

Existing customers get access to these features immediately. New customers can contact us to start a free SKE trial.

Another step towards programmable platform operations

Provisioning infrastructure is only the beginning of platform engineering. The harder challenge is operating that platform over months and years as capabilities evolve, security patches arrive, and organisations continue to grow.

Upgrade Orchestration brings those Day Two operations into the same declarative model that Kratix already applies to platform capabilities. Instead of manually coordinating platform changes, platform teams can define upgrade policies once and apply them consistently across an entire fleet.

	*Traditional Upgrade Orchestration*	With SKE Upgrade Orchestration
🛡️Safety	Upgrades affect everything at once. Mistakes ripple across your entire fleet with a huge blast radius.	Canary deployments limit the initial blast radius. Test changes on a small batch first, so mistakes stay contained.
⚡Speed	Upgrades are fast but risky, so teams delay them out of fear. Teams stay out of hours to babysit upgrades.	Batches and scheduling mean upgrades and patches roll out, your team doesn't need to worry. If something goes wrong, only the batch is impacted.
🎯Control	Limited options: upgrade everything now or manually manage each resource. No “middle ground” for a rollout No way to define scalable automated upgrade policies.	Define an upgrade policy once and reuse it as often as you need. Define more policies for more upgrade strategies. Control which resources, when, in what order, and under what conditions.
📊Reliability & Visibility	You install a version and hope for the best. No visibility into what's upgrading, progress, or what failed. Manual troubleshooting after the rollout.	See exactly what will be affected before you pull the trigger. Real-time progress tracking. Full audit trail of every upgrade run and outcome.
👥Operational Overhead	Lots of manual work. Coordinating teams, timing upgrades, monitoring, fixing problems, rolling back.	Automate routine upgrades. Focus your team on platform features, not upgrade mechanics. Scheduled jobs run while you sleep.

Every platform eventually reaches the point where operating change becomes harder than creating it. Upgrade Orchestration helps platform teams govern that change safely, consistently, and at enterprise scale.

SKE Platform Manager and Upgrade Orchestration: Available today

The SKE Platform Manager and Upgrade Orchestration are available now in the latest release of Syntasso Kratix Enterprise. If you’re an existing customer, you can get started immediately using our documentation.

Not yet an SKE customer but want to try it out? Head over to our contact form and request a trial. We'll get you set up so you can experience safe fleet upgrades firsthand.

If you're exploring how to operate internal platforms safely at enterprise scale, we'd love to show you what Upgrade Orchestration can do. My colleague, Abby Bangser, has just published a related blog post: "Walking Skeletons, Not Prototypes: Rethinking What a Platform POV Should Prove"