What Actually Happens During a Windchill 12 to 13.1 Upgrade

If you are planning a move from Windchill 12 to 13.1, the first thing worth knowing is that it is not one upgrade. It is two.

PTC's upgrade manager handles a bounded number of version steps. Once your source and target versions are far enough apart, you exceed what a single pass can do — so you run it twice. Twelve gets you to 13.0. A second pass takes 13.0 to 13.1. This is a two-hop upgrade, and it roughly doubles the effort of the version transition itself.

That surprises people, and it is usually the first schedule assumption that breaks.

What else travels with it

A version jump of that size rarely arrives alone. On a recent aerospace engagement, the same maintenance window also carried:

• An operating system upgrade underneath the application

• A vault relocation

• Re-indexing

• New replica server installations

All of it inside a single window, on a clustered environment.

As our solution architect put it, an upgrade like this is closer to open heart surgery on an engineering system than to clicking an update button. It touches nearly every component you have: the application, the web server, the runtime, the database, the operating system. And the business expects to be working on Monday morning.

Why so much of the schedule is rehearsal

Depending on the size and complexity of the environment, a Windchill upgrade of this kind runs from a three-month engagement to a year, with several people involved. Teams often assume that time is installation. It isn't. Most of it is rehearsal.

You rehearse because several of the steps cannot be undone.

Schema transformations are effectively irreversible without reverting the entire system. Vault moves carry a genuine risk of data loss, because Windchill stores information in two places — the metadata and file location live in the database, the physical file lives on disk. Get those out of sync and the damage is very hard to spot when you have millions of objects. You may not find it for months.

So teams rehearse until the process is boring. Traditionally that means one or two full dress rehearsals, because each one is expensive in human hours.

What a bad weekend looks like

The failure most people imagine — catastrophic data loss — is not the common one. The common one is more mundane and still costly.

The upgrade manager crashes partway through a database transformation. You cannot go forward, because the schema is half-migrated. So you go backward: revert the system, and lose the weekend.

Then the real cost arrives. You reschedule. You find another maintenance window that the business will accept. You re-book the people — not just the administrators, but validation, quality, and whoever else the project depends on. In a regulated environment, that can be weeks of calendar for a few hours of failed work.

That knock-on effect is why upgrade projects look disproportionately long from the outside. The work isn't slow. The cost of getting it wrong is high enough that the schedule is built around not getting it wrong.

Choosing the right approach for the vault

One decision deserves more attention than it usually gets: how you handle the vault.

Depending on circumstances you might use built-in revaulting, a manual copy, or direct SQL updates to the database tables. These are not interchangeable. Picking the wrong one can stretch a step into weeks, or — with manual database work you are not completely fluent in — create exactly the metadata-to-file desynchronisation you were trying to avoid.

There is no universal right answer. There is a right answer for your environment, and finding it before the weekend is the entire job.

Questions worth answering before you commit to a date

• How many hops does your source-to-target version gap actually require?

• What else has to move in the same window — OS, vaults, replicas, indexes?

• Is the environment clustered, and does anything in your cluster configuration predate the target version?

• How heavily customised is the system, and is every customisation compatible with 13.1?

• How quickly can you reset a test environment? This one quietly determines how many rehearsals you can afford.

• What is the rollback plan, and who signs off on invoking it?

If the answer to several of those is "we're not sure," that is not a reason to delay. It is the scope of the planning work, and it is better discovered now than at 2am on a Saturday.

Element Consulting has completed 127 PLM implementations for aerospace, defense, medical device and industrial manufacturers. If you are planning a Windchill upgrade, we're happy to talk through what it would involve in your environment.