A few days back, Oleg over at Daily PLM Think Tank did another good post titled PLM, Cloud, SaaS and Software Upgrades. It looks at several different software delivery trends (i.e. on-premise vs in-cloud) to really get at what I consider a major issue… enterprise PLM upgrades of highly customized environments.

Here are some paraphrased thoughts I left as a comment on his post with additional comments for context.

Everyone knows, upgrades can take 6-9+ months for major global PLM installations depending on the level of customization. At Aras our position is that whether SaaS or on-premise, the real issue is in the system's architecture.

If the system relies on conventional methodology where object model is hard-coded, then if customer does customization they are 'breaking' the object model. Now, they're land-locked. Next time a release comes out its incompatible with the changes made. In many cases, a highly customized global deployment effectively requires a re-implementation and data migration.

Now, if the architecture separates the object model from the code (like Aras Innovator), then you can upgrade the code without impacting the customizations (no matter how much customizing is done). It's basically an "upgrade in-place" with no data migration, no programming rewrites, etc.  Makes it fast for the engineering IT team and transparent to the user; they probably don't even know it happened.

With this type of architectural approach the deployment scenario doesn't matter: on-site, in Cloud, or hybrid deployment (i.e. vault behind firewall in on-site datacenter and application servers in cloud).

This architecture is how Aras is implemented… and what it means for companies running our PLM system is that for highly-customized global deployments at scale      upgrades for a major release take 1/100th of the man hours required for Windchill, Teamcenter, or others.

Have touched on this in a past PLM blog post titled PLM Software's Future: Model-driven vs. Model-based, but it was really focused on other advantages and not just upgrades.

Details matter, enterprise system architecture dictates a company's ability to both customize and upgrade, whether on-site or SaaS/cloud.

What's your take? Is the upgrade monster here to stay or will a better approach save us just in the nick of time?